5
Acquiring and Applying Emerging Tools and Methods

There are a host of new and emerging tools and methods that are changing the nature of environmental science and environmental health that will become increasingly important to the scientific impact of the U.S. Environmental Protection Agency (EPA) Office of Research and Development (ORD) in the years to come.

As part of identifying advanced approaches for ORD, Chapter 3 presented a One Environment–One Health framework which includes foresight for the identification of emerging challenges as well as advanced tools and methods that can help address those challenges. The advanced tools and methods can be applied within an integrated approach that considers responses to multiple stressors at biological levels ranging from molecular to individuals, populations, and ecosystems. The three challenges outlined in Chapter 3 (interconnected human health and ecological risks; environmental justice and cumulative risks; and anticipating and responding to the human health and environmental impacts of climate change) exemplify problems that can be more effectively addressed through systems-level thinking and application of new technologies. For example, the concept of one stressor leading to one effect does not account for most disease etiology, but has been the default approach to hazard and risk assessment and most regulatory approaches. Knowledge of the underlying biology of chronic diseases has increased markedly in the age of using epigenomic, genomic, and biotechnology tools that make it feasible to investigate the interaction of chemical and nonchemical stressors at basic levels of biological organization. Advancements in analytical chemistry techniques allow for discovery-driven identification of a broader range of substances to which people and ecosystems are exposed, and state-of-the-art sensor technology enables exposure assessment to be more individualized and spatially granular, instead relying on general population estimates. Greater computational power is now available to support the organization of extremely large datasets (referred to as big data) in ways that can lead to improved modeling to identify and characterize patterns, trends, and associations between stressors and effects.

This chapter identifies a number of scientific and technological advances that ORD should consider in its research planning under the topics of environmental monitors and sensors, nontargeted analysis, biotechnology, participatory research, and data science and machine learning. The committee identified and evaluated advanced tools and approaches based on the scientific literature and the committee members’ experience and expertise, their understanding of EPA’s mission and important challenges, and with consideration of the committee’s statement of task. In doing so, the committee applied general criteria in a qualitative manner that are relevant to the current state of development and use of the tools:

• Reflective of cutting-edge science, engineering, and/or innovative application of an existing ORD method;
• Potential applicability, such as
  • problem identification,
  • assessment (including screening to inform subsequent decisions about the appropriate depth of additional analyses),
  • remediation or mitigation of stressor releases or exposures,
  • information exchange outside of EPA and enhancing stakeholder involvement, and
  • decisions that cross geographic, population, ecosystem, and generational boundaries;
• Potential to facilitate a systems-thinking, One Environment–One Health approach, such as holistic assessment of a broad array of expected impacts;
• Limitations, including uncertainty in results of assessments that use the tools;
• Maturity—consideration of needed improvements for applications relevant to the work of ORD;

• Collaborations with organizations outside of ORD that could facilitate the development of the tool or method; and
• Skill sets needed within ORD to apply the tools or methods, or identify appropriate collaborative partners.

The list of tools and methods is not intended to be exhaustive, but illustrative of a range of tools and methods that ORD should consider in its research planning to support EPA’s mission. Note that this report does not give specific consideration to the range of key issues concerning New Approach Methods (NAMs) for providing the same kinds of hazard and risk assessment information that is obtained through the use of animal testing. An ongoing committee of the National Academies of Sciences, Engineering, and Medicine is reviewing the variability and relevance of existing mammalian toxicity tests, specifically when it comes to human-health risk assessment, to set expectations for the use of NAMs.

The tools and methods identified in this report are largely at a relatively advanced level of maturity to allow application in support of EPA’s mission. Many of the approaches are evolving to include additional capabilities, and some of them (e.g., some of the more sophisticated computational models) could be further enhanced by additional testing and modification. As discussed in this chapter and Chapter 4, it will be important for ORD to continue to develop and maintain advanced expertise and capabilities in broad areas, including biotechnology, exposure science, modeling, geospatial analysis and others that can be applied to many problems within EPA’s mission. As many of the largest public and environmental health problems are bigger than what is covered exclusively by EPA’s regulatory mandates, the chapters also discuss the importance of collaborations with researchers outside the agency, including academia and the private sector, and other government agencies whose mandates complement EPA’s in protecting human health and the environment. Chapter 4 also discusses communication of scientific information to the public, which—while always challenging—has become extremely difficult due to political polarization and a proliferation of misinformation and disinformation. This will require a greater emphasis on social sciences, including expertise on human behavior, societal risk perception, and communication practices.

MEASUREMENTS TO INFORM EXPOSURE AND RISK ASSESSMENTS

Humans and other organisms are exposed to stressors through contact with or consumption of food, water, soil, air, and commercial products. The process of risk assessment is instrumental to EPA and other organizations in evaluating public-health and environmental concerns regarding exposure to those stressors. Risk assessment serves as an important public-policy tool for informing regulatory and technologic decisions, setting priorities among research needs, and developing approaches for considering the costs and benefits of regulatory policies (NRC, 2009).

Emerging areas of research on the health outcomes of exposure to environmental chemicals include effects on the gut microbiome (Campana et al., 2022; NASEM, 2018a), the developing human immune system (Dietert, 2014), immunotoxicity in aquatic organisms (Desforges et al., 2016; Kataoka and Kashiwada, 2021), and the epigenetic mechanism of DNA methylation which regulates gene expression (Goodman et al., 2022; Rider and Carlsten, 2019).

Climate change is expected to alter exposures to chemical, physical, and other stressors that have the potential to produce new or exacerbated adverse impacts on human health and ecosystems likely leading to increased uncertainty in exposure and risk predictions, including climate.

Exposure information is essential for assessing, preventing, and reducing risks to human health and ecosystems. Collection of better exposure data can provide more precise information regarding risk assessments within a One Environment–One Health framework and lead to improved public health and ecosystem protection, such as better characterization of high-risk populations. Typically, discrete (or targeted) exposures are addressed one stressor at a time, with a focus on the sources of the stressor, concentrations in an environmental medium, or effects on biological systems. Accordingly, cumulative environmental exposures and their effects are poorly understood.

Recent advances, including more portable instruments and new techniques in biological and environmental monitoring, present opportunities to develop more rapid, cost-effective, and relevant exposure assessments. For example, mass-produced microelectronics and micro-electromechanical systems (MEMS) are two emerging technologies that can be used to monitor a multitude of stressors, including temperature, noise, light, volatile gases, and particulate matter. MEMS technologies can be used in mobile phones, drones, computers, and wearables. Exposure science instrumentation and analytical approaches are crucial for addressing health impacts associated with climate change.

The Concept of the Exposome

The concept of the exposome was developed for obtaining a more comprehensive description of exposure and related environmental hazards (see, e.g., Wild, 2012). The exposome comprises all chemical, physical, and biological stressor exposures; social, economic, and psychological influences; and internal physiological processes and conditions during the lifetime of an individual. Aspects of exposure to stressors and other factors—type, time (duration and frequency), concentration, or intensity—provide information that may be associated with indicators of health. Exposome-related measurements include those made within an organism (e.g., biological markers), external to an individual organism (e.g., chemical pollutants or lifestyle factors), and general factors (e.g., socioeconomic conditions). Scholz et al. (2022) discussed the exposome in the context of exposure to chemical mixtures and ecological risk assessment. Recent studies have underscored the importance of considering structural determinants in disease outcome, and identifying and exploring the important nonmolecular aspects of the exposome (Senier et al., 2017).

The comprehensive approach to understanding the exposome—encompassing humans and ecosystems—can play an important role as it further develops in taking a systems-based One Environment–One Health approach to framing ORD’s research enterprise. Gao (2021) remarked that, when risk assessments conducted within a One Health framework are based on the exposome concept, they can provide more comprehensive information and lead to more effective risk management.

Environmental Monitors and Sensors

This section describes new and emerging tools and methods that can contribute to ORD’s efforts to address the country’s environmental quality and can be used to identify and measure contaminants that impact the health of humans and ecosystems. Tools and methods are broadly grouped into the categories of air quality monitoring, water quality monitoring, remote sensing, and ecosystem monitoring.

Fixed-Site Monitors

Air Quality

EPA is a key driver in the development and implementation of air quality monitors and modeling systems such as the Community Multiscale Air Quality (CMAQ) system for pollutants regulated by the Clean Air Act.¹ Advancements made to these tools and technologies, and their predictive ability, are critical for understanding the interaction between human activities, climate, and air quality. For example, extreme events, such as heat waves and wildfires, may have a substantial influence on atmospheric chemistry and the formation of pollutants, such as airborne particulate matter. EPA’s website provides data collected from fixed-site air quality monitors, along with data visualization tools and reports on outdoor air quality from data collected at monitoring sites across the United States.²

Organic and inorganic compounds in the form of gases and in airborne particulate matter play a significant role in the formation of harmful air pollutants. While the total mass of organic and inorganic

___________________

¹ See https://www.epa.gov/cmaq.

² See https://www.epa.gov/outdoor-air-quality-data.

particulate components is more routinely measured by EPA and collaborating agencies, understanding how atmospheric chemistry changes the composition and sizes of organic components of particulate matter requires a combination of a wide array of measurement techniques. Measurements of oxidized organic molecules, such as organic peroxides (Bianchi et al., 2019; Wang et al., 2023), in both gas and particle phases are important because these species form a substantial fraction of secondary organic aerosols, which promote growth of atmospheric fine particles that can contain toxic organic compounds, such as polycyclic aromatic hydrocarbons (PAHs) (Mu et al., 2018; Shrivastava et al. 2017a). Oxidized organic species measurements are especially needed in urban communities, in outdoor and indoor environments, where the population is disproportionately affected by air pollution. Currently, however, these measurements cannot be made with low-cost sensors, because these sensors often lack useful detection limits for oxidized organics and their isomers. Deploying advanced instruments, such as aerosol mass spectrometers, chemical ionization mass spectrometers, advanced gas chromatography, and high-performance liquid chromatography techniques (Bianchi et al., 2019, Wang et al., 2023), are needed for detecting and monitoring the composition of oxidized organics mainly in urban areas that are most likely to be disproportionately affected due to climate change and socioeconomic inequalities. Research agencies, such as the National Aeronautics and Space Administration (NASA), the National Science Foundation (NSF) and the U.S. Department of Energy (DOE) in collaboration with several universities, have been conducting several measurements using mass spectrometers in addition to other state-of-the-art instruments through many aircraft- and ground-based campaigns over the years in different locations globally including the United States.³ Processes that generate these oxidized organics and ultrafine particles discovered from these measurements could be included in urban air quality models such as the CMAQ modeling system and the model outputs could be used to optimize the placement of mass spectrometers in regions that are most likely to be affected by urbanization, climate change, and air pollution.

Finding: Organic compounds in the atmosphere play a significant role in the formation of harmful air pollutants that may affect populations in urban communities disproportionately. State-of-the-art monitoring instruments are available to make needed measurements. However, deploying these expensive instruments throughout urban communities presents a challenge.

___________________

³ See https://www.arm.gov/capabilities/observatories; https://espoarchive.nasa.gov; https://csl.noaa.gov/field.html;https://www.eol.ucar.edu/all-field-programs.

Personal Monitors

Personal monitors which are used to measure inhalation exposure are typically compact and located close to the breathing zone of an individual (Munir et al., 2019). Active monitors involve the use of a small air pump to draw air through a filter or similar device. The size, noise level, and appearance of active monitors are not acceptable to some users. Passive samplers use sorption or entrapment in a diffusion tube, badge, or similar device without pumps. The collected sample is then brought to a laboratory for analysis. Lipophilic polymers are being used in lightweight passive sampling devices (e.g., silicone bands or tags) that can be worn during normal daily activities to trap airborne organic chemicals (Lin et al., 2020). Different chemical classes, including flame retardants, pesticides, PAHs, phthalates, and others have been detected and quantified in silicone bands (Hamzai et al., 2022). They can be an easy tool to implement because they are easy to deploy and they can be convenient in the study of high-risk populations such as firefighters (Poutasse et al., 2020) or in community science and participatory research initiatives. Silicone tags can also be easily worn by animals and deployed in ecosystem studies (Poutasse et al., 2019).

Key uncertainties related to the passive sampling devices include the effectiveness of the polymeric material in the sampling devices for trapping different chemicals under different environmental conditions (e.g., temperature and air flowrate), calculating environmental concentrations via partition coefficients, and validating the devices for different chemicals. The utility of passive sampling devices could be tested in populations affected by airborne hazards, such as military personnel or populations affected by wildland fires and other natural disasters (NASEM, 2020b). There are opportunities for ORD to collaborate with the National Institute of Environmental Health Sciences (NIEHS), in particular with the Human Health Exposure Analysis Resource, which provides analytical capacity for this and other environmental sampling devices.⁴ As more research is implemented, by ORD and other organizations, the purposes for which passive sampling devices are appropriate will become clearer and it is expected that their use in research will continue to increase.

Finding: The development and validation of personal sampling devices that are easily worn during normal activities provide novel opportunities for sampling some airborne pollutants in affected communities. Beyond their application to human populations, these sampling devices can also be deployed for ecosystem studies.

Recommendation 5-2: ORD should expand research into the use of personal sampling devices for assessing pollutant exposures affecting humans and other organisms. ORD could test the usefulness of passive sampling devices for assessing exposures of military personnel or civilian populations living near wildland fires. ORD can contribute to research addressing how environmental concentrations can be calculated via partition coefficients as well as validating the bands for different chemicals.

Remote Sensing

In this report, remote sensing generally refers to the use of sensor technologies on satellites or aircrafts for detecting and monitoring characteristics of Earth’s surface, including its solid crust, atmosphere, hydrosphere, and biosphere. Satellite sensors are increasingly being integrated with ground monitoring networks to monitor outdoor air quality in geographic areas where ground instruments are sparse or absent (see Appendix C). Rapid advancements are coming from new or planned sensor systems deployed by NOAA (VIIRS [Visible Infrared Imaging Radiometer Suite] and GOES-R Advanced Baseline Imager),

___________________

⁴ See https://hhearprogram.org.

the European Space Agency (TROPOMI [TROPOspheric Monitoring Instrument]), and NASA (Tropospheric Emissions: Monitoring Pollution [TEMPO] and Multi-Angle Imager for Aerosols [MAIA]). EPA plays an important role in these missions as a key supplier of ground-based calibration/validation measurements and as a consumer of final air quality products. NASA’s MAIA mission engages a multi-disciplinary science team that includes epidemiologists and health professionals and is directly targeted at improving human health. MAIA will focus on globally dispersed large metropolitan “target areas” and will be an important proof of concept for future multi-scale systems that are intended to cost-effectively integrate air quality and public health information, satellite measurements, and chemical transport models. NASA’s TEMPO mission will also provide new capabilities for monitoring transport and air quality impact of wildfire smoke, which is not well measured by current EPA regulatory monitoring networks. EPA has partnered with the U.S. Forest Service and other land management agencies to develop a fire and smoke map through a pilot project that incorporates temporary monitors and air quality sensor data to provide spatially improved air quality information and associated public health messaging during wildfire season.

The rapid evolution of small satellite (<500 kg) systems during the past two decades raises the potential for monitoring air quality variables, such as PM_2.5 (particulate matter with diameters 2.5 micrometers and smaller), at very high spatial and temporal resolution. With additional research and algorithm development, such imagery might be applied to community-scale air quality monitoring and wildfire detection and plume monitoring (le Roux et al., 2021; Tan, 2020).

Although the role of greenhouse gases (GHGs) in Earth’s energy balance—especially carbon dioxide and methane, which account for ~90 percent of present global warming—is relatively well understood, the distribution and trends in sources and sinks of those gases are less well quantified. Ground instruments combined with aircraft- and satellite-borne sensors provide important complementary estimates of GHG fluxes that could not only improve the national inventory but also inform state and local inventories and mitigation efforts (Gurney and Shepson, 2021).

In situ surface and aircraft-borne instruments provide precise, accurate “bottom-up” data on GHG concentrations and fluxes, but that measurement network is relatively sparse. An international fleet of satellite-borne sensors now provides global data on column CO₂ and CH₄ concentrations that can be assimilated with these ground measurements in atmospheric models that can then be inverted to estimate CO₂ and CH₄ budgets on a roughly 0.5 by 0.5 degree global grid. (See Appendix C.)

Finding: Remote sensing of air quality provides synoptic information that complements and can be used synergistically with existing EPA and other ground monitoring networks. Upcoming satellite missions will provide air quality information for the conterminous United States. at unprecedented spatial and temporal detail. EPA is a key partner in these missions.

Ground instruments combined with aircraft- and satellite-borne sensors provide important complementary estimates of GHG fluxes that could inform EPA’s national inventory as well as state and local inventories and mitigation efforts.

Realizing the full potential of surface monitoring networks and remote sensing data depends on their fusion and assimilation into statistical and mechanistic modeling frameworks.

Indoor Monitoring

Air quality in indoor environments where those living in developed economies spend about 90 percent of their time is not regulated (Klepeis et al., 2001). According to EPA, pollutant concentrations indoors can be higher than levels outdoors (EPA, 2021b), and the agency provides guidance for low-cost air pollution monitors for indoor air quality (IAQ) (Williams, 2019). The COVID-19 pandemic has brought attention to improving IAQ to reduce transmission of the airborne SARS-CoV2 virus (EPA, 2022b). Many are now calling for global World Health Organization IAQ guidelines, national IAQ standards, ventilation standards, and monitors for IAQ (Morawska et al., 2021). The U.S. National COVID-19 Preparedness Plan (White House, 2022) called for guidance to improve ventilation and asked that organizations participate in the “Clean Air in Buildings Challenge” launched by EPA shortly thereafter (EPA, 2022a). The challenge provides guidance to help building owners and operators improve IAQ and protect the public by creating a clean indoor air action plan, optimizing fresh-air ventilation, enhancing air filtration and cleaning, and conducting community engagement, communication, and education (EPA, 2022a).

The recent National Academies’ report Why Indoor Chemistry Matters discussed recent developments of advanced analytical techniques, such as nontargeted high-resolution mass spectrometric approaches, that have enabled the identification of many indoor contaminants (NASEM, 2022). To support further discovery, the report also presented recommendations for improvements in analytical methods and nontargeted approaches, as well as development of harmonized databases for chemical information. To obtain time-activity data for exposure and health studies, GPS and location and motion sensors embedded in smartphones and wearables have been used (Chatzidiakou et al., 2022; Duan et al., 2022). Ecological momentary assessment (EMA) is another technique that allows researchers to gather data in real time in a subject’s normal setting, as they go about their lives (Stonewall et al., 2020).

Indoor Pathogen Monitoring

Aerosols⁵ play a role in disease transmission, as they may include many smaller particles that can carry viruses, linger in the air, and accumulate indoors, increasing the potential for viral exposures. Increased potential for viral exposure is exacerbated by a changing climate, where warming temperatures and rising seas are forcing biological species to migrate and thus come into contact with animals they might not have interacted with previously.

In parallel with the well-established disciplines of aerosol transport and particle filtration, principles governing aerobiology (the study of the dispersion of airborne biological materials) are equally important but much less well developed. In the context of airborne viruses and bacteria, such airborne pathogens experience a loss of viability or ability to cause infection over time, through desiccation and exposure to atmospheric conditions. There is not much that is known about the mechanisms responsible for this loss of viability in air and how atmospheric factors and pathogen morphology mediate the loss. EPA researchers are investigating indoor air pathways of exposure and other related topics, with the goal of developing safe and effective mitigation measures to help reduce disease transmission.⁶ Other researchers are developing instruments capable of collecting airborne viruses and bacteria and maintaining their viability through the collection process (Lednicky et al., 2020; Li et al., 2022). This will allow investigators to move beyond measuring airborne concentrations of genetic material, which does not directly address disease transmission. Aerosol-levitating devices (such as the Controlled Electrodynamic Levitation and Extraction of Bio-aerosols onto a Substrate [CELEBS] instrument developed by researchers at the University of Bristol) are

___________________

⁵ Aerosols is a shorthand term for particles that are a stable suspension of solid and/or liquid particles in air, smaller than about 100 µm (NASEM, 2020a).

⁶ See https://www.epa.gov/covid19-research.

enabling a more precise—temporally, spatially, and with respect to atmospheric conditions—examination of aerosol aging and pathogen loss of viability over the very short timescales (seconds to minutes) that are most relevant to understanding and preventing human-to-human transmission of respiratory diseases (Fernandez et al., 2019; Oswin et al., 2022).

Finding: The COVID-19 pandemic has brought to light shortcomings in understanding how the spread of airborne pathogens occurs and can be mitigated. ORD researchers are involved in developing an understanding of exposure pathways for SARS-CoV-2 and reduce the risk of transmission. Other researchers are developing techniques for collecting airborne microbes and maintaining their viability, as well as examining aerosol aging and pathogen loss of viability.

Recommendation 5-4: ORD should partner with the appropriate agencies within National Institutes of Health (e.g., the National Institute of Allergy and Infectious Diseases), the U.S. Department of Homeland Security, and the Centers for Disease Control and Prevention (e.g., the National Institute for Occupational Safety and Health) to advance and expand aerobiology research (the study of the dispersion of airborne biological materials) as well as sponsor or support integrated studies that bring together engineering, medical, epidemiological, and other scientific frameworks in an effort to lessen the disruption of and strengthen defenses against seasonal infectious diseases and periodic pandemics in which significant airborne transmission occurs.

Analysis of Multiple Stressors

Water Quality

The growth in analytical capability and the standardization of water analytical methods provides a unique opportunity to characterize water contaminants and characteristics both for drinking water and natural water resources. The use of both targeted and untargeted methods can provide unique opportunities for water research, exposure and risk assessment, and policy development. Important efforts have recently started in this direction. For instance, the U.S. Geological Survey (USGS) is leading a multi-agency/institutional effort (EPA, NIEHS, the U.S. Food and Drug Administration [FDA], state and local agencies, academia) for a drinking water research collaboration focused on assessing the potential for human exposures to natural and anthropogenic contaminants in point-of-use drinking water (tap water) from private and public drinking-water sources across the nation (Bradley et al., 2018). The goal of this research is to address critical gaps in the current understanding of the extent of chemical and biological contaminant exposures via tap water and the real versus perceived importance of drinking-water contaminant exposures as drivers of adverse human health outcomes.

These research efforts in drinking water can also be applied to natural resources water studies. For instance, following similar targeted approaches measuring numerous chemicals, research has identified that chemical mixtures of anthropogenic origin including pharmaceuticals are common in U.S. streams, showing the ever-growing interconnectedness between the human and natural environment and highlighting the need to study the impact of anthropogenic chemicals on our natural environment and the active involvement of EPA in those efforts (National Water Quality Program, 2021; USGS Water Resources, 2020).

Water Quality Databases

EPA manages multiple nationwide water databases used mostly for compliance evaluation. The major drinking water database is the Safe Drinking Water Information System (SDWIS) Federal Reporting System,⁷ which includes information periodically reported by states and tribes as required by the Safe

___________________

⁷ See https://www.epa.gov/ground-water-and-drinking-water/safe-drinking-water-information-system-sdwis-federal-reporting.

Drinking Water Act (SDWA). For each public water system in the country, SDWIS contains the system’s name, ID number, city or county served, number of people served, type of system (e.g., residential), period of operation (year-round or seasonally), source water type (surface water, groundwater), violations (schedules, treatments, maximum contaminant levels), communication to customers, and enforcement (including actions taken for compliance). SDWIS has been used in research studies including studies of environmental justice and drinking water quality regarding nitrate and arsenic violations across the United States (Foster et al., 2019; Pennino et al., 2020; Schaider et al., 2019). These studies, while interesting, lack information on actual levels of exposure beyond compliance and have not been expanded to evaluate the potential health implications of exposure to drinking water contaminants.

The Six-Year Review of drinking water regulations databases are managed by EPA. They contain monitoring records that are voluntarily reported by states, territories, and tribes for contaminants in water systems regulated by SDWA. The most recent database is the Third Six-Year Review (SYR3), which includes data from 2006 to 2011 and provides the most comprehensive data on levels of water contaminants regulated by SDWA including inorganic and organic chemicals and microorganisms (EPA, 2016). Despite the richness and importance of SYR3 and the extensive quality assurance (QA) and quality control (QC) assessments prior to making it public, this database is underutilized in research, with some exceptions, mostly focused on arsenic and inorganic chemicals (Alfredo et al., 2017; Mantha et al., 2017; Nigra and Navas-Acien, 2020; Nigra et al., 2020).

In 2020, SDWIS and SYR3 were merged by water system ID in a study of inequalities in public water arsenic concentrations in community water systems across the United States (Nigra and Navas-Acien, 2020; Nigra et al., 2020). This merging provided, for the first time, an in-depth comparison of arsenic exposure levels across U.S. public water systems and allowed the study of characteristics of the water systems and counties that influence arsenic exposure levels. This is the first time that such an analysis, including maps, has been put together nationwide, although the problem of arsenic contamination in community water systems is an old problem. This SDWIS-SYR3 database was then also merged with USGS databases in 2021, allowing for a data connection between arsenic in groundwater and arsenic in community water systems nationwide, an analysis that is also a first (Spaur et al., 2021), and extended to other inorganic chemicals covered by the SDWA (e.g., barium, selenium, and uranium; Ravalli et al., 2022). ORD should play a major role in efforts whose goals are developing sustainable long-term uses of EPA drinking water databases, including their integration and maintenance, and facilitating their combination with health outcome data as well as with data related to economics, environment, social and geophysical factors, and implementation to advance drinking water research in the coming decades, including modeling spatial and temporal trends, to maximize these important data resources with huge potential for epidemiology, exposure assessment, and risk assessment.

EPA is part of the National Water Quality Monitoring Council which integrates publicly available water quality data from the USGS and more than 900 state, federal, tribal, and local agencies.⁸ This Water Quality Portal uses the Water Quality Exchange (data format to share more than 380 million water quality data records and includes EPA’s decommissioned STORET database). These data are a tremendous asset which has been incorporated into public user tools for site-specific monitoring and assessment.⁹ Though the data are extensive, there remains a significant part of ambient waters in the United States that is not monitored, or the monitoring data are sparse. Without data, the quality of ambient waters and ecosystem health cannot be assessed. Nevertheless, EPA is commended for continuing to expand and upgrade these tools and hopefully will continue to do so.

EPA is also promoting use of ambient water quality and environmental data through their “How’s My Waterway” and “BASINS” user tools.¹⁰ These are providing an important service to the public, agencies, and tribal organizations. As the need for current environmental data access increases, particularly

___________________

⁸ See https://www.waterqualitydata.us.

⁹ See https://www.epa.gov/awma.

¹⁰ See https://www.epa.gov/ceam/basins-download-and-installation; https://www.epa.gov/waterdata/hows-my-water-way.

concerning the understanding of cumulative stressor impacts and environmental restoration and conservation needs, these tools warrant expansion in their capabilities and new tools developed to include remote sensing and other databases.

EPA’s research efforts for chemicals that are still not regulated in water at the federal level are of critical importance. For instance, the widespread presence of per- and polyfluoroalkyl substances (PFAS) in drinking water and other water sources is one of the major ecological and public health concerns in the United States and other countries. These chemicals are called “forever chemicals” due to their persistence in the environment and include thousands of anthropogenic chemicals currently approved for use in numerous products. Research leadership and multi-disciplinary efforts currently ongoing by EPA (including analytical chemistry, toxicity and health effects, drinking water treatment, and cleanup and management of contaminated sites and waste; EPA, 2022c) are thus critical to inform strategies to minimize contamination and their health and ecological consequences as well as potential future legislation.

Finding: EPA has a leadership role in maintaining nationwide databases and developing analytical tools and methods for drinking water systems as well as for ambient water and ecosystem health. These databases are critical for water research to inform future regulatory action and protection measures for drinking water and ambient water quality. EPA’s research efforts for chemicals that are still not regulated in water at the federal level are of critical importance. For instance, the widespread presence of PFAS in drinking water and other water sources is one of the major ecological and public health concerns in the United States and other countries.

Recommendation 5-5: ORD should collaborate with USGS and other organizations to develop a water quality system that integrates databases for public drinking water systems, domestic wells, and ambient water from across the United States to boost water surveillance and research nationwide. The development of an integrated, curated database that is continuously maintained for water quality monitoring and research should be a major goal of EPA in the coming years. EPA’s Air Quality System, which has played a major role in boosting air quality research in the last decades, could be used as a model for this effort.

Ecosystems Monitoring

Ecosystems

Technologies for biodiversity and ecosystem monitoring are advancing rapidly along many fronts ranging from molecular methods, such as barcoding and environmental DNA (eDNA) metabarcoding, to portable biomonitoring sensors to ground-based, airborne, and satellite remote sensing (see Appendix C). Lightweight, inexpensive bio-loggers now make it possible to track animal movements, behavior, and physiology at scales not previously possible (Hellström et al., 2016; Jetz et al. 2022; Nathan et al., 2022; Smith and Pinter-Wollman, 2021). Such data can inform ecotoxicological risk assessments by revealing patterns of animal exposure as well as physiological and behavioral responses to those exposures. Similarly, low-cost bioacoustic recorders now allow cost-effective monitoring of biodiversity across a broad range of taxa, helping to track community responses to habitat loss and climate-related shifts in the timing of species’ migration and life history events.

Because EPA depends on federal and state partners for much of the information on biodiversity and ecosystem function that it uses to assess ecological risks and ecosystem services, technological advances in biomonitoring indirectly benefit EPA by improving the quality and quantity of information provided by those partners. As information on species improves, EPA and its partner agencies will be able to design mitigation efforts better tailored to specific locations, populations, and ecosystems. Biomonitoring technologies also directly support EPA’s research into species of concern and wildlife exposure to pollutants. Sensor improvements and satellite-based tracking, used in concert with inexpensive environmental

sensor systems, can be used to extend such capabilities to observe movement, behavior, and physiology and habitat conditions for many more species and environments.

As the diversity, quantity, and quality of remotely sensed data as well as computing power continue to grow exponentially, so does the range of applications to environmental assessment and monitoring (e.g., Cavender-Bares et al., 2020; de Araujo Barbosa et al., 2015; Lahoz-Monfort and Magrath, 2021; Wang and Gamon, 2019). Rapidly growing image archives, such as the Landsat Collection 2 archive and Harmonized Landsat Sentinel 2 (HLS) products, present an opportunity for EPA to develop refined indicators of ecosystem function and services that exploit the archive’s temporal depth and spatial and radiometric fidelity. Looking ahead, missions such as NASA’s Surface Biology and Geology (SBG) Deliverable and the European Space Agency’s Copernicus Hyperspectral Imaging Mission for the Environment (CHIME) will provide new hyperspectral sensor capabilities. Improved retrieval algorithms are allowing rapid gains in the ability to monitor ecosystems globally with increasing spatial, spectral, and temporal resolution. These emerging technologies can help address public health and economic issues and determine their linkages with environmental issues. Ultimately, though, realizing the full potential of these various data depends on their fusion and assimilation into statistical and mechanistic modeling frameworks. Designing and implementing these frameworks could involve multi-agency partnerships, collaboration with national ecological monitoring networks, such as the National Ecological Observatory Network (NEON), and engagement with data-model integration efforts, such as the Ecological Forecasting Initiative (EFI) as well as synthesis research centers, such as the Environmental Data Science Innovation & Inclusion Lab, the USGS John Wesley Powell Center for Analysis and Synthesis, and the National Center for Ecological Analysis and Synthesis (NCEAS).

Small satellite constellations enable local-to-global monitoring of Earth surfaces at unprecedented spatiotemporal resolution. Although they do not meet the rigorous performance standards of large national systems, lower-cost commercial systems designed to meet specific information needs can provide complementary data to the Earth observing systems discussed above for applications such as disaster management or detection and monitoring of fire, land clearing, oil spills, and pest species invasions (e.g., Lake et al., 2022).

Finding: The advent of satellite-born imaging spectrometers with high spatial and temporal resolution will enable monitoring of a wide array of One Environment–One Health issues, such as public ecosystem health indicators, landscape and water quality linkages, watershed stressors, and biodiversity of relevance to several EPA program goals.

Recommendation 5-6: ORD should investigate expanding its use of current and emerging remote sensing technologies and data products to advance its capability to model and monitor species movements, species habitats, ecosystems, and ecosystem services. In doing so, ORD should cultivate multi-agency partnerships and collaborations with research initiatives concerned with national ecological monitoring, data-model integration, and synthesis research.

Identifying Exposures Through Nontargeted Analysis

Nontargeted analysis (NTA) is a qualitative analytical technique that utilizes chromatography coupled with mass spectrometry to identify novel or unknown candidate chemicals in environmental matrices (air, water, wastewater, soil, particulate matter, and biosolids) and human biospecimens (blood, urine, feces, breastmilk, breath, hair, and nails), with applications in natural and engineered processes (i.e., bioremediation, disinfection by-products), metabolomics, pollutant exposure, toxicology assessment, and disease (Baduel et al., 2015; Li et al., 2017; Schymanski et al., 2015). The NTA approach is described in Appendix C.

New applications related to exposures in environmental and human health delve into nonquantifiable assessments, including NTA and sociological indicators, and analysis of wastewater. NTA of sample receptors (silicone bands, air-sampling backpacks), biomonitoring, and water or food explore pollutants not

previously studied or frequently monitored and provide information on broad chemical and biological exposures. Wastewater-based epidemiological studies include measurement of chemicals and microbes in community-level human excreta, including drug and contaminant metabolites, the gut microbiome, and other health indicators. Advances in DNA sequencing methods and computational analyses are enabling the use of DNA-based -omics to study microbial community structure and dynamics in drinking water (Bruno et al., 2022). Sensors for cardiovascular health, contaminants in water, air pollutants, and other indicators will provide real-time and targeted monitoring. New developments in sensors to measure specific exposure biomarkers with lower limits of detection will provide an additional data stream for individuals and researchers to assess environmental exposure and risk. The inclusion of structural determinants and health outcomes in exposure studies will allow for a more comprehensive analysis. Biomarker data have been compiled in EPA’s Exposome Explorer¹³ within the CompTox Program, summarizing health outcome information on pollutants and diet. These data can be used to determine results from chemical and biological exposures; however, drawing a link between societal factors and biomarkers is complex and requires additional expertise in the areas of environmental justice and cumulative impacts (see Chapter 3).

EPA’s ORD has developed an NTA framework for exposure and toxicity applications (EPA, 2021a; Sobus, 2016). This multi-layer approach focuses on analytical methods (Level 1), chemical candidates (Level 2), followed by toxicity analysis. To improve NTA in environmental applications, EPA has aggressively built analytical and computational capacity by initiating ENTACT (EPA’s Non-Targeted Analysis Collaborative Trial). ENTACT is a collaborative of 30 labs working on standardization of nontargeted methods (Ulrich et al., 2019). Additionally, the efforts of BP4NTA (Benchmarking and Publications for Non-Targeted Analysis), a federal, academic, and industrial working group, aim to build consensus in chemical reporting, standardize QA and QC, and make NTAs open-source data that other researchers can reference (Sauer et al., 2020).

In NTA, mass spectra are used to identify novel or unknown candidate chemicals and structural features in the original sample. The barriers of chemical identification and quantification currently limit the use of NTA in exposure analysis and health studies. Methods in human studies that link exogenous compounds with endogenous biomarkers of effect are in their infancy. However, EPA is making mass spectra easier to interpret through development of online tools and support. Future NTA needs to include improved analytical instrumentation and chemical identification, expansion and accessibility of mass spectral databases, and quantification.

Other NTAs related to environmental and human health include the -omics approaches. In metagenomics, next-generation sequencing (NGS) has improved the detection of gene products in complex systems. (See Appendix C.)

The challenges of NTA lie in

1. There is no consensus in analytical methodology. Multiple methods are required to analyze the bulk of chemicals (gas chromatography-mass spectrometry, liquid chromatography-mass spectrometry, but do not capture all chemicals.
2. Chemical identification is from generated spectra.
3. There are gaps in spectral databases supporting mass spectra analysis.

Analytical methods from lab to lab diverge in several areas: chromatographic separation (e.g., mobile and stationary phases, flow rates, column specifications), mass analysis (e.g., ionization techniques such as matrix-assisted laser desorption/ionization, electrospray ionization, mass separation), and data processing. Methods exist for targeted analytes, but the variability of sample matrices, instrumentation, and applications of NTA may prevent standardization.

The future landscape of NTA includes expanding environmental applications, novel approaches, quantifying identified or unidentified analytes, and improved identification of chemicals from mass spectra. Environmental applications that could see increased usage of NTA include odors (gas chromatographyolfactometry-mass spectrometry, olfactory sensing to detect odors) in light of odor loss or detection as a side effect of COVID-19, protein NTA to detect viruses (such as SARS-CoV-2 variants), metabolites in

wastewater-based epidemiology (targeting metabolic adducts such glucuronides, sulfates, hydroxyls), and single-cell -omics. Novel applications include combinations of NTA, such as protein-substrate chemical crosslinking, to examine active amino acid residues (proteomics and chemical NTA). Quantitative NTA is limited by current technologies, but improvements over relative quantification in mass spectrometry (e.g., fold difference and relative abundance) are possible with instrument developments and the use of chemical surrogates, internal standards, and response modeling (McCord et al., 2022). Lastly, database development will enable researchers to make more accurate identification of chemicals. Together, quantitative and qualitative NTA can prioritize chemicals in exposure and risk assessment.

Monitoring and Evaluating Exposures

Finding: Local-scale and personal monitoring is becoming increasingly sophisticated, enabling more comprehensive assessment of exposure. Sensors for cardiovascular health, water contaminants, air pollutants, and other indicators will provide real-time and targeted monitoring. New developments in sensors to measure specific exposure biomarkers with lower limits of detection will provide an additional data stream for individuals and researchers to assess environmental exposure and risk. New applications are including the use of NTA, which involves the use of qualitative analytical techniques to provide information on unstudied or infrequently monitored pollutants in environmental media (air, water, and soil) and human biospecimens (e.g., blood and urine).

Recommendation 5-7: ORD should expand its use of available data on stressor exposures collected from local-scale and personal exposure monitors to improve exposure and risk assessment for humans and ecosystems. This includes partnering with other agencies and nongovernmental organizations that fund and have expertise in researching and managing projects providing air and water exposure measurements for terrestrial and aquatic biota and humans at a community level. ORD should enhance its efforts to apply nontargeted analyses to such data to ensure that the broadest range of possible exposures of concern can be identified.

An Integrated Database

Finding: Over the course of briefings from EPA and our own experience, it was remarkable how many large-scale datasets have been used or could be used to improve exposure and risk assessment and modeling of global phenomena. Large-scale observations of air quality, temperature, GHG concentrations, river flows, bio-logging of wildlife movements, temporal trends in LandSat data, and others were identified as having provided information valuable for addressing research and regulatory questions. In this era of big data, EPA has the opportunity to gain insights into air and water quality, chemical exposures and hazards, and cumulative risk in ways not previously possible.

Recommendation 5-8: ORD should play a major role in developing integrated databases with collaborating organizations, such as the U.S. Department of Agriculture (USDA), NOAA, and USGS, for water, air, and soil quality from across the United States; pollutant toxicity; exposure to humans and other organisms; human health data, and other parameters that are mission relevant. In this era of big data, EPA has the opportunity to gain insights into environmental chemical exposures and hazards and cumulative risk in ways not previously possible. However, development of this integrated database will require considerable planning for and implementation of capabilities that ORD is beginning to develop but will need to bring to the next level of expertise and capacity.

BIOTECHNOLOGY

For the purposes of this discussion, we consider “biotechnology” to refer to methods, products, and processes other than selective breeding and sexually crossing organisms to endow new characteristics in

organisms (NASEM, 2017). It usually involves the introduction of or change in genetic information of plants, animals, or microorganisms for making or altering products or processes. This section considers examples of advanced tools and methods for evaluating potential human health and ecosystem impacts of biotechnology applications (e.g., consumer products and energy generation) as well as an example of an advanced tool to support EPA’s role in environmental protection. Additional information is provided in Appendix D.

Increasingly, biotechnology applications are moving outside of containment and into the natural and built environments. For example, genetically modified crops have been used for years, and applications for disease vector control are being designed to spread or persist within the open environment. Many of these biotechnology products have unique properties, and the environmental pathways through which possible human health and ecological harms could occur are often uncertain. Biotechnologies are expected to have further impacts on society as applications increase across multiple areas while costs and barriers to access fall. Both the speed of change and breadth of the applications make the task of forecasting the impacts of biotechnologies both urgent and difficult.

EPA, FDA, and USDA have primary responsibility for ensuring the safety of biotechnology and the commercial-scale production of biological molecules and materials (biomanufacturing). EPA is responsible for biotechnology products that have insecticidal, fungicidal, rodenticidal, or other toxic properties. It also has authority over new chemicals in commerce that include certain forms of genetically engineered organisms.

ORD, in turn, faces the challenge of providing effective approaches for human health and ecological risk assessments for planned open releases into the environment and monitoring to detect unintended impacts of biotechnology products. A One Environment–One Health approach would provide a framework for an integrated consideration at a molecular level to understand the engineered genes and resulting phenotypic characteristics of the host organisms and at higher levels to carry out ecological and human health risk assessments of possible hazardous outcomes.

Genetically modified organisms (GMOs) are increasingly used in biotechnology for more efficient, targeted biotransformation processes. Concern over introducing modified cells into the environment arises regarding the potential for the genetic modification to transfer to other microbiota in the environment, for the mutant strain to overtake the native population during the specific biotech application, or for the strain to lose the introduced gene. Therefore, reliable methods are needed to identify the intact host strain and the modified gene for monitoring the GMO’s distribution in the environment, as well as the occurrence of the modified genetic materials outside of the host organism.

Routinely used methods for identifying the presence of GMOs in a soil or water sample can be time-consuming. Single-molecule sequencing (Eid et al., 2009) is a method that can vastly improve resolution and output in determining the order of nucleotides collected from a sample. Single-molecule sequencing has many applications in biotechnology, microbial surveillance, and detection of genetic polymorphisms.

Single-cell sequencing examines information from individual cells, providing a higher resolution of cellular differences compared with traditional sequencing technology (Nawy, 2014; Shapiro et al., 2013; Tang et al., 2019; Wang and Navin, 2015). The technique has applications in health monitoring, genomics, and transcriptomics. However, methods to sequence a single strain’s 16S rDNA, chromosome, and plasmid for taxonomic identification have yet to be developed, but, when available, the method will enable detection of GMOs in a mixed consortia collected from the environment.

Other areas of biotechnology research are examining how to increase the potential precision with which changes can be made in plant genomes and expand the number of characteristics that can be changed or introduced (NASEM, 2016). These advances have shown the potential for designing plants with new and enhanced traits (Medford and McCarthy, 2017), enabling them to be used as biosensors (Bick et al., 2017; Singh et al., 2020)—to detect pollutants, for example—and in water purification applications. They can also be applied in ways that harness the plant as a “toolbox,” using it to produce other types of outputs, such as chemical precursors for vaccines (Reed et al., 2017). Many of these applications are still in the early research and development stages, which may impact their applicability to ORD.

Emerging biotechnologies represent compounding scientific, societal, and governance challenges. ORD, and EPA more broadly, will need the resources and expertise related to the evaluation and utilization of emerging biotechnologies to address environmental issues such as climate change, systems-level implications of these applications, and how to examine these applications against a broader set of societal questions such as those related to environmental justice and public engagement. Consideration of these broader implications will reach beyond ORD and EPA and point to the need for collaborations with other federal agencies as well as organizations outside of government. There may be opportunities set forth in Executive Order 14081, the White House Initiative to Advance the Bioeconomy, to help facilitate such collaborations (Biden, 2022).

Finding and Recommendation

Finding: Biotechnology continues to evolve rapidly and it will likely be used for a variety of purposes that may include altering microorganisms to metabolize specific contaminants in waste sites, creating novel plants and organisms for a variety of purposes, including applications designed for the built and natural environment and many more uses that fall under both EPA’s regulatory purview and potential research areas within ORD. Advanced tools are becoming available for monitoring the distribution of genetically modified microorganisms in the open environment and the occurrence of the modified genetic materials outside of the host organism.

Recommendation 5-9: ORD should strengthen and maintain its expertise in biotechnologies to evaluate and monitor the potential human health and ecosystem impacts of new biotechnology applications in a One Environment–One Health framework, as well as develop research opportunities for EPA to utilize these tools for environmental protection applications, such as waste remediation and pollution monitoring. In addition, ORD should

• Build partnerships with other federal agencies and conduct periodic horizon scanning activities to stay abreast of emerging biotechnologies and applications that those agencies may be funding.
• Collaborate early and often with researchers and risk assessors to design relevant ecological and human health risk assessments and interpret the results.

PARTICIPATORY RESEARCH APPROACHES

The history of participatory research methods is connected to the environmental justice movement, which arose as a fusion of Indigenous environmental rights, the civil rights movement, environmentalism, and public health protection in response to environmental racism (Wilson, 2010). The term participatory research can be utilized across an array of practices, and thus has different meanings to different people. For reference, the participatory research practice can also be referred to as citizen science, community science, non-establishment research, and public participation in science. The main distinction between participatory research and traditional (or establishment) research is that participatory researchers may not necessarily have obtained formal research training from accredited institutions (Rasmussen et al., 2020). Participatory research broadens the scope of who can contribute to scientific initiatives and can be a pathway for obtaining new information and understanding (NASEM, 2018b). Participatory research provides a means for local residents to play a substantial role in identifying potential environmental concerns (such as hotspots or trends of increased pollution) and considering alternative risk mitigation approaches.

A community-driven method, such as participatory research, is critical for collecting data outside the reach of current investigatory systems and methods and is critical for better understanding of environmental justice and cumulative risk, with the recognition that community input and engagement are essential for both investigation and mitigation strategies. Community-engaged research comprises diverse approaches in the fields of environmental health science, social sciences, and other disciplines that aim to

democratize the scientific enterprise in ways that enhance environmental health literacy and transform research from an expert-driven process into one of co-learning and co-production of knowledge (Balazs and Morello-Frosch, 2013).

Through its involvement in participatory science EPA has developed collaborations with communities addressing environmental justice concerns and initiatives with states and tribal sovereign tribal nations. Examples of the uses of participatory research by EPA and partner organizations include assisting with emergency response actions, evaluating new air-quality-sensors, assessing water quality, and contributing to environmental education (EPA, 2022d). EPA developed a StoryMap to highlight participatory research projects to support environmental monitoring.¹¹

As EPA continues to embrace participatory research, communities that are gaining access to emerging technologies and developing research mechanisms and protocols and data validation tools can illustrate ways in which ORD can harness the energy and talents of these communities.

For example, do-it-yourself biology is a global movement spreading the use of biotechnology and synthetic biology tools beyond traditional academic and industrial institutions into community-based laboratories and education centers (see Appendix E). Another community-based science and technology initiative is the FabLab network, whose mission is to provide access to tools, knowledge, and financial means for education, innovation, and invention using technology and fabrication.¹²

Building relationships and using a participatory approach, with the community at the center, are key components of research activities with Indigenous communities. In recent years, a growing number of Indigenous scientists are calling for increased leadership and authority of Indigenous organizations in science. An example is the Native BioData consortium, a research institute led by Indigenous scientists and tribal members in the United States.

Indigenous communities in the United States have developed research codes and methods to organize and regulate the collection and circulation of information gathered about their members, for the purpose of protecting themselves from potential abuses in biomedical research. These codes have been successfully used by longstanding participatory research studies, such as the Strong Heart Study¹³ and the Navajo Birth Cohort Study,¹⁴ conducted in partnerships between Indigenous communities and academic institutions (see Appendix E). These studies and the research codes that have made them possible are at the forefront of developing ethical review processes and participatory research models to protect individuals and communities from the unintended consequences of research. The principles and research codes include

• Traditional ecological knowledge to promote understanding of relationships between people and their environment based on accumulated observations transmitted through generations.
• Collective leadership to ensure fully engaged participatory research and trust.
• Sovereign data ownership so that investigators and funding agencies acknowledge that data collected on tribal nation members are owned by the nation—not by the investigator, academic institution, or the federal government—and, as such, there are a series of codes and procedures that the investigators need to follow to access and use those data.
• Specific review processes for studies, protocols, publications, and granting authority to share data and transfer materials. For instance, many tribal communities have set in place research review boards to review the research activities.
• The 7th generation principle to consider the downstream impact of decision-making on multiple generations.
• Relationality, which involves acknowledgment that things are connected in a continuous circular process of growth rather than in a linear process. This principle of relationality is helpful

___________________

¹¹ See https://storymaps.arcgis.com/stories/57b2ee78221341a18b0f7ebe8017340d.

¹² See https://fabfoundation.org.

¹³ See https://strongheartstudy.org.

¹⁴ See https://hsc.unm.edu/pharmacy/research/areas/nbcs-echo.html.

• within the context of systems thinking, where it is likely that numerous feedback loops and adjustments in multiple processes need to be considered.

Finding and Recommendation

Finding: Participatory research methods and the environmental justice movement have come to the central stage in environmental health sciences, with the recognition that community input and engagement are essential to address the major environmental injustice challenges affecting society, including investigation and mitigation strategies. Indigenous communities in the United States have developed a series of research codes and tools to regulate the collection and circulation of information about their members.

DATA SCIENCE AND MACHINE LEARNING

Recent advances in artificial intelligence, machine learning, and data science hold the promise of providing an array of research tools for supporting EPA’s mission.¹⁵ Advances in computational power have enabled the aggregation and manipulation of massive datasets (see Box 5-1) to create sophisticated models for estimating physical, chemical, and biological phenomena. For example, leveraging big data can improve the ability to identify previously undetected links between exposure to stressors and health outcomes (e.g., causal associations between exposures to low concentrations of PM_2.5 and human mortality) (see Box 5-2), and to estimate future trends.

___________________

¹⁵Artificial intelligence refers to the science and engineering of developing computers and other machines with the ability to imitate human problem-solving and decision-making capabilities. Machine learning refers to the development and use of computer systems that gradually improve the accuracy of their output by imitating humans learning. Data science is an interdisciplinary field that involves the use of scientific methods and algorithms to extract insights from data (structured or unstructured).

This section discusses modeling and data science advances in assessing chemical toxicity, influences of air quality on biological air contaminants, and the potential effectiveness of pollution control scenarios. It also discusses the kinds of workforce expertise and infrastructure ORD needs to position itself to take advantage of those advances in the coming years.

Data and Models for Assessing Chemical Toxicity

Traditional chemical toxicity testing protocols, which mainly rely on animal models, are expensive and time-consuming (Luechtefeld et al., 2018; Meigs et al., 2018; Zhu et al., 2014). Existing computational models for chemical toxicity evaluations include quantitative structure–activity relationship (QSAR) modeling approaches (see Appendix F).

ORD is currently active in modeling related to toxicology and risk assessment. For example, ExpoCast provides exposure estimates for chemicals in commerce based on their presence in consumer products and environmental media and compares the results to biomonitoring data from National Health and Nutrition Examination Survey and other sources (Wambaugh et al., 2013). Those estimates provide a basis for prioritizing chemicals based on potential for high exposure and, combined with high-throughput screening (HTS) of biological activity from ToxCast (Dix et al., 2007; Judson et al., 2010; Kavlock et al., 2012) and the associated Tox21 program (Hsu et al., 2017; Shukla et al., 2010; Thomas et al., 2018), can be used as a basis for prioritization for further testing and regulatory action. HTS for biological activity provides a basis for prioritizing chemicals for more in-depth assessment.

In addition, the Virtual Embryo program has used cell-agent–based modeling to predict key cellular events in the development of embryonic organs and structures and estimate how they can be perturbed by environmental agents. Computational modeling of the development of blood vessels was successful in identifying adverse effects of potent inhibitors of angiogenesis, a result that was confirmed by wet lab models (Kleinstreuer et al., 2013). The model also boosts confidence in the ability of HTS methods to estimate vascular developmental toxicity for less-potent agents (Saili et al., 2019).

Recent advances in applications for assessing chemical toxicity include:

• Big Data¹⁶Applications: Advances in experimental protocols (especially HTS), and combinatorial chemistry (Attene-Ramos et al., 2013; Hsu et al., 2017; Inglese et al., 2006; Lehn, 1999; Malo et al., 2006) have resulted in the generation of a huge amount of biological data for millions of compounds (NRC, 2007). In addition, data-sharing projects make massive data on chemicals publicly available (Ciallella and Zhu, 2019; Hartung, 2016; Luechtefeld et al., 2018; Zhao and Zhu, 2018; Zhu et al., 2014, 2016). In addition, data-sharing projects such as PubChem make massive data on chemicals publicly available (Wang et al., 2009, 2010). As a

___________________

¹⁶ The term “big data” refers to structured or unstructured data that are so large and multi-faceted as to be impossible to evaluate using traditional computational approaches.

• result, these advances provide a foundation for developing innovative modeling approaches to fulfill the needs of chemical risk assessments.
• High-Throughput Exposure Modeling: This can complement HTS to identify chemicals that have predicted human exposure levels in a concentration range that elicits biological activity. This approach has been refined by incorporating high-resolution mass spectrometric data to increase the number of chemicals identified in environmental mixtures (Rager et al., 2016). Wambaugh et al. (2013) described a model for predicting exposure to a large number of chemicals using a combination of data about use patterns (e.g., personal care products and fragrance) and biomonitoring data to predict a range of exposure values, which could then be compared to biological activity data (Wetmore et al., 2015). This research, which originated within ORD, is an excellent example of combining existing expertise in risk and exposure assessment with sophisticated modeling and analytical chemistry.
• Data-Driven Modeling: Analysis of big data requires the use of advanced tools, such as heterogeneous and cloud computing and dynamic data curation and sharing with algorithms that handle data streams (Charikar et al., 2003; Liu et al., 2007). These advanced techniques allow for rapid identification of target entities in massive datasets, which has radical implications for the improvement of traditional computational toxicology modeling techniques, such as a method for estimating toxicological endpoint information for a substance by using data on the same endpoint from another substance (read-across) (Guo et al., 2019; Hartung, 2016; Russo et al., 2019; Zhao et al., 2020b). Recent HTS programs and their associated data-sharing efforts have revolutionized approaches used in many health fields (see Appendix F).

  The advancement of data science technology and the development of data-driven computational tools also aim to overcome challenges posed by big data in computational toxicological modeling: volume (amount of data), velocity (growth of data), variety (diversity of sources), and veracity (uncertainty of data). Those challenges create new opportunities for existing toxicity model improvement and novel model development (Andreu-Perez et al., 2015; Ciallella and Zhu, 2019; Lee and Yoon, 2017). The datasets available for chemical toxicity modeling, such as those from ToxCast and Tox21, involve compounds tested against many targets (Thomas et al., 2018). Traditional QSAR modeling and machine learning approaches are not always suitable for this type of data. This challenge, along with data quality questions, is a major obstacle in using big data (Ciallella and Zhu, 2019; Zhao et al., 2020a; Zhu, 2020). This big data application needs to be coupled with the development of new computational approaches to deal with high-volume, multi-dimensional, and high-sparsity data sources to predict chemical toxicity (Zhang et al., 2017; Zhu, 2020).

• Mechanistic Modeling: Guidance from the Organisation for Economic Co-operation and Development calls for the predictions of computational models to have mechanistic explanations (OECD, 2018). Most HTS assays were developed to query specific biological targets, allowing for mechanistic interpretation (Krewski et al., 2010). The large amount of mechanistically explainable data available enables researchers to create computational models that incorporate adverse outcome pathway concepts (Ankley et al., 2010) (see Appendix F). The resulting models can be defined as mechanistic models that predict the toxicity of new compounds and elucidate mechanisms (Ciallella and Zhu, 2019; Wittwehr et al., 2017; Zhu et al., 2014, 2016).

Identifying Influences of Air Quality on Biological Air Contaminants

The emergence of COVID-19 as a global pandemic instigated perhaps unprecedented demand for expertise in aerosols and indoor air quality (see Appendix F). The intense public interest in health risks posed by biological air contaminants points to the need for ORD to leverage the atmospheric chemical kinetic modeling expertise of its scientific staff and partners, in collaboration with other federal agencies, to build greater understanding of the chemical kinetics in aerobiology (the study of the dispersion of airborne biological materials). Put into practice, such understanding could reveal how infrastructure, structural

determinants, and personal choices drive indoor air composition, which in turn may have an accelerating or retarding effect on the natural rates of decay of pathogen viability in air.

Models for Assessing Pollution Control Scenarios

Energy production and use are changing in many areas around the world with the modernization of electricity production from coal to natural gas and renewables, and a shift in transportation vehicles from gasoline to electric vehicles. There will also be an associated change in the size distribution and chemical composition of airborne particulate matter that is relevant to human health and climate in many parts of the world, including the United States. These changes are expected to vary across regions, sectors, and time. The risks attributable to climate change and extreme events (heat waves, droughts, floods, pests, wildfires, cyclones, etc.) will intersect with and compound health risks, such as those associated with the COVID-19 pandemic (Phillips et al., 2020). In a rapidly changing world, interdisciplinary cross-sectoral risk assessment tools are needed that account for key feedback loops affecting vulnerable populations, for example, related to the food–energy–water–health–infrastructure nexus.

Despite the recent progress in advancing computational models, there are still limitations for implementation. For example, three-dimensional chemical transport models (3D CTM; e.g., the CMAQ model used by EPA) estimate how air pollutants respond to changes in atmospheric chemistry and meteorology. Advanced measurements and modeling techniques could help reduce uncertainties related to process-level understanding of complex interactions between natural and human systems governing particulate matter in different CTMs (Shrivastava et al., 2017b). However, 3D CTMs are computationally expensive (Brasseur and Jacob, 2017), making it difficult to rapidly assess the role of different policy scenarios on air quality and human health. Building computationally efficient machine learning models, which emulate 3D CTM predictions, could rapidly predict the nonlinear responses of air pollutants, such as ozone and particulate matter, to precursor emissions and meteorology changes (Xing et al., 2022).

However, even the most advanced integrated assessment models fail to capture the complex nonlinear interactions across human and natural systems and across different spatial and temporal scales. Advances in mathematical algorithms, artificial intelligence, and machine learning can lead to the development of universal integrating tools for assessing multiple and unknown interactions across complex systems, with increased predictability and computational efficiency (see Appendix F). Integrated machine learning approaches have the potential to project monetary costs of policy actions that affect known and unknown stressors related to climate change, extreme weather, pandemics (such as COVID-19), and changes in built and natural ecosystems, as well as the disproportionate impacts of those stressors across socioeconomically disparate communities with additional concerns related to environmental justice and human perceptions.

Model Maturity and Limitations

Despite the tremendous progress made in developing computational models, traditional experiments are needed to assess the validity of individual models and allow for modifications to be made in an iterative fashion, whereby the modeling informs the types of experiments that need to be done, and the experimental data are used to refine the models (see Appendix F).

Workforce Expertise

As artificial intelligence and machine learning solutions are increasingly being adopted in data analytics and computational modeling, it is important to keep in mind the saying that “There is no AI (artificial intelligence) without IA (information architecture).” Federal agencies, such as ORD, generate or have access to vast datasets from many sources that are used by a network of researchers and other experts from public and private sectors, including academic institutions. Because democratizing data is essential to maximizing their value, data management capabilities for discovery, governance, curation, and orchestration comprise an important skill set for ORD. Those capabilities are needed for when ORD is making data in its

possession available or advocating for open sources of data controlled by other organizations, so that the data are ready to support applications, analytics, and process automation.

The ability to readily derive insights from data in support of anticipating and responding to environmental problems is another critical skill set for ORD so that it can take full advantage of artificial intelligence and machine learning technologies and strengthen and leverage partnerships with other organizations, such as NASA, that are directly relevant to the role ORD plays in supporting EPA’s mission.

Findings and Recommendations

Finding: Trained machine learning algorithms can be used as integrating tools for rapid assessment tools for decision-makers. For example, machine learning approaches could be used to rapidly assess the response of multi-pollutant mixtures in the atmosphere to air quality management scenarios. Novel algorithms and tools could be developed for rapid predictions of the complex interactions between human systems and natural systems, and for assessing the compound climate–pandemic health risks, including among low socioeconomic status communities.

Recommendation 5-11: To improve modeling of the complex interactions across human and natural systems and across different spatial and temporal scales, ORD should utilize recent advances in causal approaches and machine learning models, and leverage the machine learning capabilities of existing supercomputing facilities at government agencies, universities, and other organizations. In addition, it should continue to develop in-house computational, analytical, and data storage capabilities for managing diverse datasets. ORD should foster the development of collaborative machine learning frameworks that train algorithms to integrate diverse datasets rapidly for use in assessing combined risks from different stressors.

Finding: Recent advances in artificial intelligence, machine learning, and data science hold the promise of providing an array of research tools for supporting EPA’s mission and point to the need for requisite expertise and infrastructure.

Recommendation 5-12: ORD should review and identify data management infrastructure needs. It should maintain and enhance expertise enabling it to take full advantage of data management technologies as well as strengthening and leveraging partnerships with organizations across the network of state and local agencies, researchers, and other experts from public and private sectors, including academic institutions.

SUMMARY OF THE CHAPTER

This chapter identifies a number of scientific and technological advances that ORD should consider in its research planning under the topics of environmental monitors and sensors, NTA, biotechnology, participatory research, and data science and machine learning. The committee identified and evaluated advanced tools and approaches, based on the scientific literature and committee members’ experience and expertise, along with their understanding of EPA’s mission and important challenges, and with consideration of the committee’s statement of task.

The list of tools and methods is not intended to be exhaustive or define an exclusive set of highest-priority tools and approaches. Also, many of the tools and methods discussed in this chapter are evolving, and many of the largest public and environmental health problems are bigger than what is covered exclusively by EPA’s regulatory mandate. As discussed in this chapter and Chapter 4, it will be important for ORD to continue to develop and maintain advanced expertise and capabilities in broad areas, including biotechnology, exposure science, modeling, geospatial analysis, and others that can be applied to many problems within EPA’s mission.

REFERENCES

across the USA, 2006-11: A cross-sectional study. The Lancet Planetary Health 6(4):e320-e330. https://doi.org/10.1016/S2542-5196(22)00043-2.

Reed, J., M. J. Stephenson, K. Miettinen, B. Brouwer, A. Leveau, P. Brett, R. J. M. Goss, A. Goossens, M. A. O’Connell, and A. Osbourn. 2017. A translational synthetic biology platform for rapid access to gram-scale quantities of novel drug-like molecules. Metabolic Engineering 42:185-193. https://doi.org/10.1016/j.ymben.2017.06.012.

Rider, C. F., and C. Carlsten. 2019. Air pollution and DNA methylation: Effects of exposure in humans. Clinical Epigenetics 11(1):131. https://doi.org/10.1186/s13148-019-0713-2.

Russo, D. P., J. Strickland, A. L. Karmaus, W. Wang, S. Shende, T. Hartung, L. M. Aleksunes, and H. Zhu. 2019. Nonanimal models for acute toxicity evaluations: Applying data-driven profiling and read-across. Environmental Health Perspectives 127(4):1-14. https://doi.org/10.1289/EHP3614.

Saili, K. S., J. A. Franzosa, N. C. Baker, R. G. Ellis-Hutchings, R. S. Settivari, E. W. Carney, R. M. Spencer, T. J. Zurlinden, N. C. Kleinstreuer, S. Li, M. Xia, and T. B. Knudsen. 2019. Systems modeling of developmental vascular toxicity. Current Opinion in Toxicology 15:55-63. https://doi.org/10.1016/j.cotox.2019.04.004.

Sauer, U. G., R. A. Barter, R. A. Becker, E. Benfenati, E. Berggren, B. Hubesch, H. M. Hollnagel, K. Inawaka, A. M. Keene, P. Mayer, K. Plotzke, R. Skoglund, and O. Albert. 2020. 21st century approaches for evaluating exposures, biological activity, and risks of complex substances: Workshop highlights. Regulatory Toxicology and Pharmacology 111:104583. https://doi.org/10.1016/j.yrtph.2020.104583.

Schaider, L. A., L. Swetschinski, C. Campbell, and R. A. Rudel. 2019. Environmental justice and drinking water quality: Are there socioeconomic disparities in nitrate levels in U.S. drinking water? Environmental Health 18(1):3. https://doi.org/10.1186/s12940-018-0442-6.

Scholz, S., J. W. Nichols, B. I. Escher, G. T. Ankley, R. Altenburger, B. Blackwell, W. Brack, L. Burkhard, T. W. Collette, J. A. Doering, D. Ekman, K. Fay, F. Fischer, J. Hackermüller, J. C. Hoffman, C. Lai, D. Leuthold, D. Martinovic-Weigelt, T. Reemtsma, N. Pollesch, A. Schroeder, G. Schüürmann, and M. von Bergen. 2022. The eco-exposome concept: Supporting an integrated assessment of mixtures of environmental chemicals. Environmental Toxicology Chemistry 41(1):30-45. https://doi.org/10.1002/etc.5242.

Schymanski, E. L., H. P. Singer, J. Slobodnik, I. M. Ipolyi, P. Oswald, M. Krauss, T. Schulze, P. Haglund, T. Letzel, S. Grosse, N. S. Thomaidis, A. Bletsou, C. Zwiener, M. Ibáñez, T. Portolés, R. de Boer, M. J. Reid, M. Onghena, U. Kunkel, W. Schulz, A. Guillon, N. Noyon, G. Leroy, P. Bados, S. Bogialli, D. Stipaničev, P. Rostkowski and J. Hollender. 2015. Non-target screening with high-resolution mass spectrometry: Critical review using a collaborative trial on water analysis. Analytical and Bioanalytical Chemistry 407(21):6237-6255.

Senier, L., P. Brown, S. Shostak, and B. Hanna. 2017. The socio-exposome: Advancing exposure science and environmental justice in a postgenomic era. Environmental Sociology 3(2):107-121. https://doi.org/10.1080/23251042.2016.1220848.

Shapiro, E., T. Biezuner, and S. Linnarsson. 2013. Single-cell sequencing-based technologies will revolutionize whole-organism science. Nature Reviews Genetics 14(9):618-630. https://doi.org/10.1038/nrg3542.

Shrivastava, M., S. Lou, A. Zelenyuk, R. C. Easter, R. A. Corley, B. D. Thrall, P. J. Rasch, J. D. Fast, S. L. M. Simonich, H. Shen, and S. Tao. 2017a. Global long-range transport and lung cancer risk from polycyclic aromatic hydrocarbons shielded by coatings of organic aerosol. Proceedings of the National Academy of Sciences of the United States of America 114:1246-1251, https://doi.org/10.1073/pnas.1618475114.

Shrivastava, M., C. D. Cappa, J. Fan, A. H. Goldstein, A. B. Guenther, J. L. Jimenez, C. Kuang, A. Laskin, S. T. Martin, N. L. Ng, T. Petaja, J. R. Pierce, P. J. Rasch, P. Roldin, J. H. Seinfeld, J. Shilling, J. N. Smith, J. A. Thornton, R. Volkamer, J. Wang, D. R. Worsnop, R. A. Zaveri, A. Zelenyuk, and Q. Zhang. 2017b. Recent advances in understanding secondary organic aerosol: Implications for

global climate forcing. Reviews of Geophysics 55(2):509-559. https://doi.org/10.1002/2016RG000540.

Shukla, S. J., R. Huang, C. P. Austin, and M. Xia. 2010. The future of toxicity testing: A focus on in vitro methods using a quantitative high-throughput screening platform. Drug Discovery 15(23-24):997-1007. https://doi.org/10.1016/j.drudis.2010.07.007.

Singh, S., V. Kumar, D. S. Dhanjal, S. Datta, R. Prasad, and J. Singh. 2020. Biological biosensors for monitoring and diagnosis. Pp. 317-335 in Microbial Biotechnology: Basic Research and Applications, J. Singh, A. Vyas, S. Wang, and R. Prasad (eds.). Springer Nature Singapore. https://doi.org/10.1007/978-981-15-2817-0_14.

Smith, J. E., and N. Pinter-Wollman. 2021. Observing the unwatchable: Integrating automated sensing, naturalistic observations and animal social network analysis in the age of big data. Journal of Animal Ecology 90(1):62-75. https://doi.org/10.1111/1365-2656.13362.

Sobus, J. 2016. Advancing Non-Targeted Analysis Research Within EPA/ORD. U.S. Environmental Protection Agency Office of Research and Development. https://www.epa.gov/chemical-research/advancing-non-targeted-analysis-research.

Spaur, M., M. A. Lombard, J. D. Ayotte, D. E. Harvey, B. C. Bostick, S. N. Chillrud, A. Navas-Acien, and A. E. Nigra. 2021.Associations between private well water and community water supply arsenic concentrations in the conterminous United States. Science of the Total Environment 787:147555. https://doi.org/10.1016/j.scitotenv.2021.147555.

Stonewall, J., M. Dorneich, U. Passe, and L. Shenk. 2020. Implications for using mobile ecological momentary assessment to collect climate data from vulnerable urban populations. Proceedings of the Human Factors and Ergonomics Society Annual Meeting 64(1):273-277. https://doi.org/10.1177/1071181320641064.

Tan, S. 2020. Remote sensing applications and innovations via small satellite constellations. Pp. 767-797 in Handbook of Small Satellites, J. N. Pelton (ed.). Springer Nature Switzerland. https://doi.org/10.1007/978-3-030-36308-6_44.

Tang, X., Y. Huang, J. Lei, H. Luo, and X. Zhu. 2019. The single-cell sequencing: New developments and medical applications. Cell & Bioscience 9(1):1-9. https://doi.org/10.1186/s13578-019-0314-y.

Thomas, R. S., R. S. Paules, A. Simeonov, S. C. Fitzpatrick, K. M. Crofton, W. M. Casey, and D. L. Mendrick. 2018. The US federal Tox21 program: A strategic and operational plan for continued leadership. ALTEX: Alternatives to Animal Experimentation 35(2):163-168. https://doi.org/10.14573/altex.1803011.

Ulrich, E. M., J. R. Sobus, C. M. Grulke, A. M. Richard, S. R. Newton, M. J. Strynar, K. Mansouri, and A. J. Williams. 2019. EPA’s non-targeted analysis collaborative trial (ENTACT): Genesis, design, and initial findings. Analytical and Bioanalytical Chemistry 411(4):853-866. https://doi.org/10.1007/s00216-018-1435-6.

USGS (U.S. Geological Survey) Water Resources. 2020. Down the Drain? Pharmaceutical Presence a Concern for Aquatic Life in Some Streams. February 5. https://www.usgs.gov/news/pharmaceuticals-common-small-streams-us.

Wambaugh, J. F., R. W. Setzer, D. M. Reif, S. Gangwal, J. Mitchell-Blackwood, J. Arnot, A. O. Joliet, A. Frame, J. Rabinowitz, and T. B. Knudsen. 2013. High-throughput models for exposure-based chemical prioritization in the ExpoCast project. Environmental Science & Technology 47(15):8479-8488. https://doi.org/10.1021/es400482g.

Wang, R., and J. A. Gamon. 2019. Remote sensing of terrestrial plant biodiversity. Remote Sensing of Environment 23:111218. https://doi.org/10.1016/j.rse.2019.111218.

Wang, Y., and N. E. Navin. 2015. Advances and applications of single-cell sequencing technologies. Molecular Cell 58(4):598-609. https://doi.org/10.1016/j.molcel.2015.05.005.

Wang, Y., J. Xiao, T. O. Suzek, J. Zhang, J. Wang, and S. H. Bryant. 2009. PubChem: A public information system for analyzing bioactivities of small molecules. Nucleic Acids Research 37:W623-W633. https://doi.org/10.1093/nar/gkp456.