Submission for new seed suggestions is currently closed.
 | Example abstracted from one of our PREEMPT Phase 1 Institute - Improving our ability to predict outbreaks of Spruce budworms in northeast american forests Nina Fefferman | ||||||||||||||||||||||||||||||||||||||||||||
Description
Spruce budworm outbreaks devastate fir and spruce forests in the northern U.S. and southern Canada. The last big outbreak in the east started in the 1970s, and a new incipient outbreak is being suppressed currently in s.e. Canada. Spruce budworms follow a general pattern of outbreaks every 40-ish years, and there are data available on outbreak-vs. -not for about 300 years. There are multiple mechanistic and phenomenological models of outbreaks of this species, but none that are effective predictors of outbreaks. One proposed model predicts alternate stable states for budworm populations (outbreak & low numbers), but they might be phase shifts caused by changed environmental conditions. Shifts in inter-outbreak intervals could be affected by nationwide declines in bird populations (they eat caterpillars), climate change, and human actions (wood demand & suppression). We could delve into the problems of budworm outbreak modeling to see if we can come up with an effective predictive model
Deliverables
a predictive model with code implementation; a published academic journal manuscript; a whitepaper targeting governmental and industry managers and policy makers to help with sustainable forestry practices
Policy
improved predictions may be able to improve active management strategies for both public and private forests
Disciplines or Expertise Required for this Working Group
Comments:
| Report for Moderation Links Spruce Budworm Benefit-cost analysis of spruce budworm (Choristoneura fumiferana Clem.) control: Incorporating mark Files | ||||||||||||||||||||||||||||||||||||||||||||
| Theorectical model explaining patterns of rotavirus in children and adults Alanna Hoyer-Leitzel | ||||||||||||||||||||||||||||||||||||||||||||
Description
Rotavirus is a gastrointestinal infection that presents in children, but rarely in adults. There are multiple theories for this pattern of infection. We would like to propose a theoretical model in which a low-dose, regular exposure to the virus can lead to long term immunity but with a transient period of reoccurring infections. Previous work (see the linked paper and attached supplemental figure) using a model parameterized for influenza A shows a similar pattern. This model could be re-parameterized for Rotavirus, potentially giving a theoretical explanation for patterns of Rotavirus infection in children and adults.
Deliverables
manuscript published in an academic journal
Policy
possible recommendations on interventions to prevent infections
Disciplines or Expertise Required for this Working Group
Comments: | Report for Moderation Links Influenza A paper Files appex_supplement_rotavirus.pdf
| ||||||||||||||||||||||||||||||||||||||||||||
| Looking for statistical early warning signals of disease outbreaks Michael Reed | ||||||||||||||||||||||||||||||||||||||||||||
Description
Select several diseases where time series of data are available, then use a variety of statistical approaches to determine if there are early warning signals (EWS) of outbreaks. We would select diseases with very different transmission characteristics in order to look for commonalities in EWS. I think this could be done with a small team of interested folks, particularly if one of them was a postdoc who could dedicate time to data wrangling and analyses.
We might also look into potential non-disease indicators that could correlate with outbreaks, such as Google trends, or social media indicators.
Deliverables
At least 1-2 applied papers, and perhaps a methodological paper as well.
Policy
If EWS exist, it allows the possibilities fo anticipation and preparation for outbreaks; even better, it might allow intervention to prevent outbreaks.
Disciplines or Expertise Required for this Working Group
Comments: | Report for Moderation Links data source for highly pathogenic avian influenza data source for highly pathogenic avian influenza data source for RSV data source for ebola data source for C. auris Files | ||||||||||||||||||||||||||||||||||||||||||||
| Cultivating Social & Health Resilience, in Response to Variable Disease Threats Alexander Pritchard | ||||||||||||||||||||||||||||||||||||||||||||
Description
Humans have been posited as adaptable or versatile in response to variable circumstances. Such adaptability has been supported through evolutionary perspectives of responses to instability. This perspective parallels concepts of physiological or social resilience, whereby individuals or groups, respectively, express a capacity to maintain homeostasis in response to stress. Disease presents a whole-group stressor with, potentially, unequal consequences for individuals in the group. From this simple perspective, variable responses will emerge and provide multiple solutions to risk. Yet, social groups must arrive at a consensus agreement that generally aligns, to maintain cohesion. Thus, social perspectives of group resilience are potentially at odds with incorporating variable responses to risk. How do humans prioritize variable solutions to disease risk, and its related stressors, while maintaining group cohesion?
Deliverables
Academic publications; Organizational outcomes (i.e., cultivating resilient working groups); Policy (indirect outcomes, e.g., fostering care groups in healthcare)
Policy
(No direct policy implications are immediately apparent)
Disciplines or Expertise Required for this Working Group
Comments: | Report for Moderation Links Files | ||||||||||||||||||||||||||||||||||||||||||||
| Is Lassa virus reassorting, and if so where?: Implications for surveillance and control Mete K. Yuksel, Courtney L. Schreiner | ||||||||||||||||||||||||||||||||||||||||||||
Description
Recombination and reassortment, processes that shuffle genetic variation in viruses, have been implicated in the evolution, spillover, and emergence of zoonotic diseases such as influenza and SARS-CoV-2. But there is disagreement about the frequency and importance of these processes. We propose using Lassa virus, a rodent-borne RNA virus that is endemic in West Africa, to develop a better understanding of how often –– and where –– viral genomes may be recombining or reassorting. To do this, we will use publicly-available sequence data to estimate reassortment rates and networks for Lassa in different parts of its range. If Lassa is reassorting, we will apply phylogeographic methods to estimate the speed of Lassa dispersal and pinpoint the locations of genetic ancestors. This knowledge could be used to identify where novel strains of Lassa may arise and, thus, inform surveillance and control efforts.
Deliverables
Manuscript, recommendations for sequence collection and surveillance.
Policy
Understanding of where to allocate resources to combat Lassa virus spillover and emergence.
Disciplines or Expertise Required for this Working Group
Comments: | Report for Moderation Links https://www.annualreviews.org/content/journals/10.1146/annurev-virology-101416-041726 https://academic.oup.com/mbe/article/39/2/msac028/6519867 https://www.pnas.org/doi/10.1073/pnas.1521582113?url_ver=Z39.88-2003&rfr_id=ori%3Arid%3Acrossref.org&rfr_dat=cr_pub++0pubmed https://www.nature.com/articles/s41579-022-00789-8 https://lassa.nkn.uidaho.edu/ https://www.pnas.org/doi/10.1073/pnas.1918304117 https://elifesciences.org/articles/72177 Files | ||||||||||||||||||||||||||||||||||||||||||||
| Wastewater-based cases/active infections model Pablo Paiewonsky, Nancy Grisell Ramirez Herrera (Quantum Risk Analytics, Inc.) | ||||||||||||||||||||||||||||||||||||||||||||
Description
In a burgeoning epidemic or pandemic, knowledge of new cases and/or active infections (NCAI) is needed since these are the most immediate indicators of upcoming hospitalizations and deaths as well as how infection is spreading. NCAI are almost always imperfectly observed due to lack of testing and, potentially later on, underreporting of test results. Methods for estimating NCAI from reported cases and test positivity rates have already been developed (e.g., Chiu and Ndeffo-Mbah, 2021) as well as from hospitalization and death data. However, such signals usually lag behind wastewater viral concentrations. The latter can be translated into NCAI via calibration with case estimates like the aforementioned, perhaps modulated by environmental variables. We propose developing such a technique, as well as exploring the ability of spatial interpolation methods (e.g., kriging) to extend wastewater-based NCAI estimates to locations without wastewater viral sampling.
Deliverables
1 or 2 manuscripts on methodology and results, conference presentation of the work, open-source-available computer code which process the relevant data and estimate NCAI progression based on wastewater prevalence data and geographical location.
Policy
Wastewater surveillance can provide the latest information about active infections in a region, which can then be better used to inform intervention measures and short term forecasting and also allocate extra resources to health care providers. Moreover, NCAI estimates from wastewater surveillance can be used to complement other such estimates, especially when severe cases are few or when other monitoring has ceased.
Disciplines or Expertise Required for this Working Group
Comments: | Report for Moderation Links Chiu and Ndeffo-Mbah, 2021 Li et al., 2023 Rauch et al., 2024 Mohring et al., 2024 Files | ||||||||||||||||||||||||||||||||||||||||||||
| Framework for Pandemic Modeling: a Modular Infrastructure for Multi-Scale model integration Richard D. Hamlin (Quantum Risk Analytics, Inc.) | ||||||||||||||||||||||||||||||||||||||||||||
Description
We propose a modular stochastic framework for pandemic modeling that integrates in-house or third-party submodels via a Software Development Kit and API for multi-scale disease transmission simulations, supporting compartmental, agent-based, and mechanistic models. Our framework enables researchers to test assumptions, compare models, and leverage probabilistic programming (Pyro) for Bayesian inference without requiring submodels to adopt Pyro. A Macro Model enhances classical compartmental modeling with Bayesian inference, geographic stratification, and, via coupling factors, dynamic mobility, regional interactions, and demographic risks. It also integrates Micro Models (e.g., indoor airborne transmission) and Vaccine Effectiveness Models. By standardizing inputs and outputs, our framework boosts collaboration, scalability, and computational efficiency, enabling real-time scenario testing with mobility, vaccination, and crowdsourced data for pandemic forecasting and risk assessment.
Deliverables
Academic publications, conference presentations, whitepapers, new tools and methodologies, advanced simulations, optimization of vaccination strategies, collaboration, real-time data integration, improved epidemiological models, flexibility and scalability. Milestones: prototype demonstration, new models/modules (phylogenetic, behavioral, CFD) and validation, early policy simulation case studies, beta testing with base datasets/models, user surveys, grant awards, and initial framework adoption.
Policy
Governments could use our framework to design better (more nuanced, targeted, adaptive, effective, and economically efficient) public health policies, and vaccination strategies based on real-time data, providing better risk assessments for WHO and CDC, and could use it to optimize hospital capacity, vaccine distribution and emergency response planning.
Disciplines or Expertise Required for this Working Group
Comments: | Report for Moderation Links IDM Symposium Poster Presentation (highlighting the Framework and some key components/features) Whitepaper: Macro Model (which forms the core of the Framework) / Features Compared Whitepaper: Vaccine Effectiveness Model (Probabilistic model generates VE models used in Framework) High-Level Model/Framework Diagram (with captions) Pandemonium project website Pandemonium project and app early draft proposal Pandemonium Open-source GitLab repos (small sampling of our code; much more to be added) Interview: COVID-19 Risk Assessment App Pandemonium Uses AnyChart for Data Visualization Files | ||||||||||||||||||||||||||||||||||||||||||||
| Adaptive Socioeconomic-Epidemiological Model for Optimal Pandemic Response Richard D. Hamlin, Nancy Grisell Ramirez Herrera (Quantum Risk Analytics, Inc.) | ||||||||||||||||||||||||||||||||||||||||||||
Description
Pandemics cause cascading crises in health, economy and society. Traditional models either focus on epidemiology or assume economic equilibrium, failing to capture nonlinear disruptions. We propose a model that integrates nested agent-population structures with supply chain networks to stimulate economic and behavioral responses dynamically.
Hybrid Epidemiological-Economic Framework: Uses Representative Agent Groups to scale behaviors efficiently to macro levels.
Supply Chain Network Modeling: Captures disruptions, bottlenecks and cascading failures.
Behavioral Economics integration: Models panic buying, labor force shifts and adaptive firms strategies.
Real-time policy adaptation: Combines Bayesian inference (updating model parameters for evolving observed pandemic data) & Reinforcement Learning (optimizing policy given prior policy effectiveness).
This approach enhances pandemic response strategies by adapting interventions in real-time for better economic resilience & public health.
Deliverables
Peer-reviewed studies demonstrating the model´s effectiveness in pandemic response planning, conference presentations, whitepapers, early input from stakeholders, interactive tools for policymakers, frameworks to analyze and optimize public health and economic interventions, partnerships with governments, prototype demonstration, model validation, early policy simulation case studies, beta testing with base datasets/models, user surveys, grant awards, model adoption.
Policy
Adaptive health policies leveraging real-time behavioral and economic data, strategies for economic resilience and resource optimization.
Disciplines or Expertise Required for this Working Group
Comments: | Report for Moderation Links Adaptive Socioeconomic-Epidemiological Model for Optimal Pandemic Response, presentation Pandemonium project website Files | ||||||||||||||||||||||||||||||||||||||||||||
| Mapping across scale: from clinically phenotyped cohorts to larger populations of infected individuals: part 1 Profs. Lydia Bourouiba and Sanjat Kanjilal | ||||||||||||||||||||||||||||||||||||||||||||
Description
Detailed phenotyping of clinical cohorts in infectious diseases can typically only be done in comparatively small number of patients (~ hundreds at best), though the clinical phenotyping can be done in detail and very precisely (e.g., including clinical surveillance). A critical challenge is how to map such cohorts of patients, associated with detailed physiological and clinical metrics, to a larger population dynamics and less granular metrics of surveillance. This project will explore such methodological mapping approaches in a rational and rigorous manner leveraging existing databases and surveillance systems for infectious diseases in clinical settings.
Deliverables
Mathematically rigorous approach to mapping individuals of a larger cohort to individuals in a highly phenotyped set of patients.
Policy
Policy on surveillance of emerging infectious diseases and mapping of community to care facility emergence of infectious disease outbreaks would be guided/enabled.
Disciplines or Expertise Required for this Working Group
Comments: | Report for Moderation Links https://www.populationmedicine.org/skanjilal https://lbourouiba.mit.edu/ Files | ||||||||||||||||||||||||||||||||||||||||||||
| Mapping across scale: from clinically phenotyped cohorts to larger populations of infected individuals: part 2 Profs. Lydia Bourouiba and Sanjat Kanjilal | ||||||||||||||||||||||||||||||||||||||||||||
Description
Detailed phenotyping of clinical cohorts in infectious diseases can typically only be done in comparatively small number of patients (~ hundreds at best), though the clinical phenotyping can be done in detail and very precisely (e.g., including clinical surveillance). Another critical challenge is how does one select from the larger cohort those patients, based on clinical and surveillance indices, who are closest to patients that have been extensively phenotyped, possibly by performing additional measurements, or obtaining additional metrics on the larger cohort and determining which measurements or metrics are the most informative for such mapping. This project will explore such methodological mapping approaches in a rational and rigorous manner leveraging existing databases and surveillance systems for infectious diseases in clinical settings.
Deliverables
Mathematically rigorous approach to mapping individuals of a larger cohort to individuals in a highly phenotyped set of patients.
Policy
Policy on surveillance of emerging infectious diseases and mapping of community to care facility emergence of infectious disease outbreaks would be guided/enabled.
Disciplines or Expertise Required for this Working Group
Comments: | Report for Moderation Links https://lbourouiba.mit.edu/ https://www.populationmedicine.org/skanjilal Files | ||||||||||||||||||||||||||||||||||||||||||||
| Adaptive machine-learning tool for computationally efficient approximations for airborne pathogen transmission where spatial and temporal heterogeneities matter, selecting & learning from hybrid CFD if needed Richard D. Hamlin, Pablo Paiewonsky (Quantum Risk Analytics, Inc.) | ||||||||||||||||||||||||||||||||||||||||||||
Description
Current models in use often favor computationally-reducing assumptions such as well-mixed-room and/or steady-state flow and viral concentrations. However, in many real world situations, conditions deviate greatly from these ideals, suggesting the use of precise transient computational fluid dynamics (CFD)-based plus agent-based modeling of susceptible and infectious humans. Because these methods combined would be computationally very demanding, we ought to know under which kinds of scenarios temporally- and/or spatially-averaged transmissions in realistic, complex models deviate substantially from those in simpler models. Some work (e.g., Zhang et al., 2021) suggests that spatio-temporal heterogeneities may be significant. If so, then hybrid simple-CFD modeling approaches may be useful, with their results extended using machine learning techniques. ML is used both in training agents from video of human movements in a space and for estimating corrections and uncertainties.
Deliverables
Systematic comparison of transmission models for key scenarios (hospitals, schools, buses, offices, stores, etc.) and modeling contexts (aggregation of indoor transmission events, risk modeling for individuals or important agents); derived neural network architecture and weights; a tool that can estimate the correction and uncertainty using ML; open-source code of our hybrid CFD model with potentially an automated mesh generator; paper(s) explaining the methodologies, validations, and results.
Policy
Make many more time- and place-specific personal risk assessments computationally feasible, thus giving a more accurate nuanced understanding of public risk.
Improved role- and location-specific infection probability accuracies (e.g., nurses versus patients in a hospital) would inform workplace interventions.
For any transmission modeling context, give inaccuracy and uncertainty levels for both simple and ML-extended models, which could obviate the need for computationally expensive simulations.
Disciplines or Expertise Required for this Working Group
Comments: | Report for Moderation Links Pandemonium project website Pandemonium project draft proposal Machine Learning / CFD model proposal Pandemonium Open-source GitLab repos (small sampling of our code; much more to be added) Salmenjoki et al. (2021) Zhang et al. (2021) Liu et al. (2021) Tan et al. (2022) Files | ||||||||||||||||||||||||||||||||||||||||||||