Summary of the NSF PIPP-PREP Workshop

Christine K Johnson, Urbashi Mitra, Nian Sun, Brian Wood

Summary and justification for research

The NSF Predictive Intelligence for Pandemic Prevention (PIPP) Workshop – Pandemic Readiness for Emerging Pathogens (PREP) was held on Feb. 16 and 17, 2021, in a virtual online format. This workshop, the first in a series of four PIPP workshops, focused on highlighting the challenges and potential solutions to rapid detection and assessment of emerging pathogens through advanced biosensors, surveillance, modern data analytics such as machine learning, and the tracking of human and animal populations for risk modeling and pandemic preparedness. This interdisciplinary workshop had >750 participants from academia, government, industry and non-governmental organizations. Research disciplines were drawn from the purview of the following four NSF directorates: Engineering (ENG), Biological Sciences (BIO), Computer and Information Science and Engineering (CISE), and Social, Behavioral and Economic Sciences (SBE). This workshop brought together specialists who are advancing rapid pathogen detection, biosensing, machine learning, and the tracking of human and animal populations for the purpose of risk modeling and pandemic preparedness.

This workshop aimed to broaden participation across a wide range of disciplines, as the scientific challenges for pandemic prevention are highly cross-disciplinary, and future pandemic readiness for emerging pathogens requires cross-disciplinary research, trans-disciplinary communication, and integrative innovation.

To address pandemic and emerging challenges, we identified the need for (a) a holistic approach to pandemic prevention that more effectively integrates disciplines; (b) design of novel educational and training programs to ensure a properly informed workforce for cross-disciplinary research; (c) development of evaluation strategies that foster high-risk/high-payoff approaches and allow for a longer timescale for achieving research results; (d) development of evaluation strategies for team-science and cross-disciplinary team-science; and (e) design of assessment strategies for determining the efficacy of translation from research to public policy.

The PREP workshop fostered scientific discussion of the challenges and potential solutions to rapid detection and assessment of emerging pathogens and infectious disease dynamics from the molecular to the ecological scale. The overall workshop objective was to enable key interaction and collaboration among the greater research community, among those who are currently engaging in needed interdisciplinary research, as well as experts in strategic areas who may not be currently collaborating across disciplines. The workshop focused on both strategic research directions as well as processes to enable key collaboration and identify methods by which high-risk, high-payoff research can be recognized and rewarded.

We developed four topical themes based on key pairings of disparate disciplines for which integration is essential for research in pandemic preparedness. The PIPP PREP workshop four topical thrusts were 1) Rapid and Accurate Detection and Assessment of Emerging Pathogens, 2) Monitoring Environmental Change, Animal Movements, and High-Risk Interfaces for Disease Transmission, 3) Monitoring Human Movements and At-risk Communities for Disease Transmission and Spread, and 4) Data-Intensive Machine Learning and Modeling for Pandemic Preparedness. Each theme had two keynote speakers as well as a panel of experts who shared strategic research directions. We endeavored to engage researchers at all stages of their careers.

Challenges, needs, and gaps for the four workshop themes

Discussion amongst workshop participants identified the following highly interdisciplinary challenges/needs/gaps for addressing the pandemic prevention within each of the four themes.

1: Rapid and Accurate Detection and Assessment of Emerging Pathogens

New sensor modalities for rapid (~seconds) and autonomous detection of pathogens in diverse environments.
New methods for improved sensor accuracy.
Innovative applications of sensing technologies that are broadly useful across taxa and locations to accurately detect and track pathogens in key wildlife populations that are likely reservoirs for emerging infectious diseases with pandemic potential.
Machine learning and artificial intelligence for improved accuracy of sensors.
Wireless sensor networks for detecting emerging pathogens in remote areas.
Data analytics for mutation prediction.
Translation of laboratory based detection methods to in-the-field rapid methods.

2: Monitoring Environmental Change, Animal Movements & High-Risk Interfaces for Disease Transmission

Design of autonomous sensing platforms for large scale animal population monitoring.
Analyzing the effect of deforestation, climate change and human habitation on the evolution and adaptation of emerging viruses.
Design of novel sensor modalities targeting animal populations to detect animal movements and changes in animal disease state.
Next generation techniques in environmental data collection with more frequent and precise measurements of landscape change and land use.

3: Monitoring Human Movements and At-risk Communities for Disease Transmission and Spread

Study the accuracy, reliability, and ecological validity of sensor-aided measures of human movement, patterns of social contact, and risks for infectious disease transmission.
Continuing work on wearable sensor technology with modest form factors for long term data collection.
Study how to improve user adoption of contact tracing technologies and increase their effectiveness for lowering disease exposure risks.
Theory driven social science and human ecology investigations into the underlying reasons why spatiotemporal contact patterns vary, and why spillover and disease exposure risks vary across social and cultural settings.
Creation of new platforms, protocols, data repositories, and data formats for sharing human movement and social contact data to help catalyze fundamental research.
Investigate biases that arise owing to differences in the rates of technological adoption (e.g. contact tracing technologies) and explore methods for improving public cooperation and buy-in.

4: Data-Intensive Machine Learning and Modeling for Pandemic Preparedness

Data challenges: initial sparsity/data scarcity at the beginning of the pandemic; optimal tradeoffs between accuracy and latency and translating these tradeoffs into interventions;
Design and/or acquisition of training sets; incorporating fairness and equity into machine learning methods; determining the optimal granularity of needed data
Novel imputation methods for missing or erroneous data; time-varying learning that adapts to evolving behavioral patterns.
Machine learning based policy design: how to mutually educate policy makers and scientists to enable co-design of data-informed policies; design of human-in-the-loop methods for data analytics; ensuring privacy and security for data collection and analysis.

Cross-disciplinary research and call for innovative integration

The scientific challenges for pandemic prevention are highly cross-disciplinary, ranging from epidemiology, virology, biology, engineering, computer and information science, social, behavioral and economic sciences, etc. Highly cross-disciplinary teams are needed to address these challenges, which include:

Holistic approaches to pandemic prevention that cover all four themes listed above are needed.
Design of novel educational and training programs to ensure a properly informed work-force/researchers for cross-disciplinary research.
Development of evaluation strategies that foster high-risk/high-payoff approaches and allow for a longer timescale for achieving research results.
Design of assessment strategies for determining the efficacy of translation from research to public policy.

In the past, emerging infectious disease research has largely been driven by the biological sciences community. Greater involvement and integration of the engineering, social science, and data science research communities is needed to identify and implement solutions with better capacity for monitoring and rapid detection as we enter the next frontier of pandemic preparedness research and advance capabilities in rapid identification and prediction.

In an outbreak or a pandemic, rapid detection of pathogens is of paramount importance and plays a key role in efforts to contain and mitigate the outbreak. Early detection saves lives, reduces person-to-person transmission, and can mitigate broader impacts on public health, livelihoods, and the economy. The future of pandemic preparedness lies in the accurate prediction of risk to animal and human populations and enhanced biosurveillance.

Pandemics by nature cross national borders and oceans. To increase the scale of pandemic prevention research to face this global challenge, US-based researchers must address the gaps in sustainable data collection and communication to enable predicting and monitoring at wildlife-human interfaces in international rural settings as well as urbanizing settings where community transmission can facilitate larger scale epidemics.

Workshop PIs (in alphabetical order)

Christine K Johnson

Professor of Epidemiology and Ecosystem Health, Director of EpiCenter for Disease Dynamics, University of California, Davis

Urbashi Mitra

Gordon S. Marshall Chair in Engineering

Professor, Ming Hsieh Department of Electrical & Computer Engineering, Department of Computer Science, University of Southern California

Nian Sun

Professor of Electrical and Computer Engineering and in affiliation with BioEngineering and Chemical Engineering

Director of the W.M. Keck Laboratory for Integrated Ferroics, Northeastern University

Brian Wood

Associate Professor of Anthropology, University of California, Los Angeles

External Scientist, Max Planck Institute for Evolutionary Anthropology, Department of Human Behavior Ecology and Culture

NSF Workshop on Predictive Intelligence for Pandemic Prevention