APPEX Seeds – APPEX

Mapping across scale: from clinically phenotyped cohorts to larger populations of infected individuals

Profs. Lydia Bourouiba and Sanjat Kanjilal

Selected

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 implications

Policy on surveillance of emerging infectious diseases and mapping of community to care facility emergence of infectious disease outbreaks would be guided/enabled.

Disciplines / expertise required

Clinical infectious diseases
Surveillance
Mathematical epidemiology
Multiscale modeling
Risk of infectious disease transmission
Mathematical epidemiology

Comments

No comments yet.

Is Lassa virus reassorting, and if so where?: Implications for surveillance and control

Mete K. Yuksel, Courtney L. Schreiner

Selected

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 implications

Understanding of where to allocate resources to combat Lassa virus spillover and emergence.

Disciplines / expertise required

Mathematical biology
Virology
Population genetics
Genomics
Geography
Statistics
Public health policy

Comments

No comments yet.

Theorectical model explaining patterns of rotavirus in children and adults

Alanna Hoyer-Leitzel

Selected

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 implications

possible recommendations on interventions to prevent infections

Disciplines / expertise required

Rotavirus
Data Processing
Parameter Estimation
Dynamical Systems Modeling
Stochastic Systems Modeling

Comments

No comments yet.

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 implications

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 / expertise required

mechanical engineer (fluid dynamicist)
biophysicist
applied mathematician
data scientist/machine learning developer
applied statistician
lead software/model developer

Comments

No comments yet.

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 implications

Policy on surveillance of emerging infectious diseases and mapping of community to care facility emergence of infectious disease outbreaks would be guided/enabled.

Disciplines / expertise required

Clinical infectious diseases
Mathematical epidemiology
Multiscale modeling
Surveillance
Risk of outbreak in healthcare

Comments

No comments yet.

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 implications

Adaptive health policies leveraging real-time behavioral and economic data, strategies for economic resilience and resource optimization.

Disciplines / expertise required

Lead Model/Software Developer Architectural Oversight: Ensures the framework is scalable, modular, and integrates various models (Macro, Micro, VE, etc.) efficiently. Guides system integration and interoperability. Cross-Team Coordination: Connects researchers, developers, and public health experts to align computational models with epidemiological objectives. Manages software and data infrastructure to support framework users including epidemiologists and policymakers. Quality Assurance & Security: Maintains compliance with security and data privacy standards. Implements rigorous code review and testing strategies to ensure reliability and accurate models. Performance Optimization: Identifies and resolves computational bottlenecks to enhance efficiency on existing hardware, analyzes trade-offs between computational costs, accuracy, and scalability. Strategic Direction: Defines technical roadmap for adaptability and sustainability, spearheads integration of emerging technologies like probabilistic programming.
Epidemiology Epidemiologists are crucial for understanding disease dynamics, transmission patterns, and population health impacts, and evaluating health-related datasets. Their expertise guides the development of accurate models and simulations within our framework. We are actively seeking professionals in this field; for more details, please refer to our Epidemiologist position: https://www.pandemonium.dev/jobs/?id=Epi
Economics Economics is essential for understanding and modeling the macro- and microeconomic effects of pandemics. The model needs to account for labor market disruptions, productivity losses, fiscal and monetary policies, and economic recovery trajectories. Key Contributions to the Model include: Representative Agent Groups—Simulating heterogeneous economic behavior (e.g., essential vs. non-essential workers). Pandemic-Driven Labor & Production Shocks—Linking epidemiological changes to workforce disruptions and economic resilience. Policy Optimization—Designing intervention strategies that balance economic and health trade-offs (e.g., lockdown vs. open economy scenarios).
Behavioral Science/Psychology Experts in human behavior and psychology will guide the development of models that predict how people are likely to respond to public health policies, disease outbreaks, and other stressors. Understanding psychological drivers is key to modeling behavior accurately. Pandemic outcomes depend on public compliance, risk perception, and adaptive behaviors. Behavioral science models panic buying, vaccine hesitancy, and work-from-home trends, all of which affect both disease spread and economic recovery. Social network analysis helps track misinformation and trust in policies.
Data Science and Machine Learning, including Probabilistic Programming (Pyro) Data scientists and machine learning experts analyze complex datasets to identify patterns and make predictions. Their skills are vital for developing probabilistic models and integrating diverse data sources into our framework. We have openings for Machine Learning Developers & Data Scientists: https://www.pandemonium.dev/jobs/?id=ML
Software Development (Python) Proficiency in Python is essential for building and maintaining the framework's infrastructure. Developers will implement algorithms, manage data processing pipelines, and ensure the software's scalability and efficiency. We are looking for Senior Python Developers: https://www.pandemonium.dev/jobs/?id=SrPy
Quality Assurance (QA) Engineering (Python) A QA Engineer ensures the reliability, accuracy, and security of the pandemic modeling framework by developing and maintaining automated and manual testing processes. Their role is critical in validating epidemiological models, software components, and data integrity. https://www.pandemonium.dev/jobs/?id=PyQA
Human Mobility and Spatial Modeling Experts in this field analyze movement patterns and spatial data to simulate how diseases spread across different regions. Their work informs the development of dynamic coupling factors in our macro model. We are seeking a Human-Mobility Spatial Modeler: https://www.pandemonium.dev/jobs/?id=Mob
Public Health Policy Professionals in this area assess the implications of model outcomes on public health and guide policy recommendations. Their expertise ensures that the framework's outputs are relevant and actionable for decision-makers.
Supply Chain Management A pandemic hurts supply chains, trade networks, and industrial production. Supply Chain Management expertise is needed to model bottlenecks, cascading failures, and logistics constraints in real-time. Cascading Failures Across Industries: A labor shortage in one sector can ripple through multiple supply chains, leading to critical shortages and inflation. Resilience & Recovery Planning: Governments and businesses need real-time insights on how to secure essential goods, adjust logistics & prioritize resource allocation. Dependencies: The pandemic response is shaped by international trade disruptions, shipping delays, and production shifts. Key Model Contributions: Supply Chain Network Modeling—Simulating production, distribution, & inventory disruptions in real-time. Labor-Driven Logistics Constraints—Assessing how labor shortages affect shipping, manufacturing & retail. Policy Integration—Informing decisions on strategic stockpiling, trade restrictions & industrial prioritization.

Comments

No comments yet.

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 implications

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 / expertise required

Lead Model/Software Developer Architectural Oversight: Ensures the framework is scalable, modular, and integrates various models (Macro, Micro, VE, etc.) efficiently. Guides system integration and interoperability. Cross-Team Coordination: Connects researchers, developers, and public health experts to align computational models with epidemiological objectives. Manages software and data infrastructure to support framework users including epidemiologists and policymakers. Quality Assurance & Security: Maintains compliance with security and data privacy standards. Implements rigorous code review and testing strategies to ensure reliability and accurate models. Performance Optimization: Identifies and resolves computational bottlenecks to enhance efficiency on existing hardware, analyzes trade-offs between computational costs, accuracy, and scalability. Strategic Direction: Defines technical roadmap for adaptability and sustainability, spearheads integration of emerging technologies like probabilistic programming.
Epidemiology Epidemiologists are crucial for understanding disease dynamics, transmission patterns, and population health impacts, and evaluating health-related datasets. Their expertise guides the development of accurate models and simulations within our framework. We are actively seeking professionals in this field; for more details, please refer to our Epidemiologist position: https://www.pandemonium.dev/jobs/?id=Epi
Data Curation Data Curators specialize in the acquisition, management, and preservation of data. This role is vital for ensuring the quality and accessibility of data within the framework. While epidemiologists will handle most health-related data curation, other Data Curators will focus on broader data management tasks. We have opportunities for Data Analysts & Curators: https://www.pandemonium.dev/jobs/?id=DA
Data Science and Machine Learning, including Probabilistic Programming (Pyro) Data scientists and machine learning experts analyze complex datasets to identify patterns and make predictions. Their skills are vital for developing probabilistic models and integrating diverse data sources into our framework. We have openings for Machine Learning Developers & Data Scientists: https://www.pandemonium.dev/jobs/?id=ML
Statistics and Biostatistics Statisticians and biostatisticians are essential for developing the Generalized Risk Factor module and ensuring that all statistical methods within the framework are robust. Their expertise in statistical theory and methods will guide the accurate interpretation of data and the reliability of model predictions. They will work closely with epidemiologists and data scientists to refine data analyses and ensure statistical integrity across the framework.
Database Development Database Developers focus on the technical aspects of database management systems, including design, implementation, and maintenance. Their work is crucial for ensuring that the framework’s underlying databases are robust, performant, and scalable. We have opportunities for Database Developers. https://www.pandemonium.dev/jobs/?id=DB
Software Development (Python) Proficiency in Python is essential for building and maintaining the framework's infrastructure. Developers will implement algorithms, manage data processing pipelines, and ensure the software's scalability and efficiency. We are looking for Senior Python Developers: https://www.pandemonium.dev/jobs/?id=SrPy
Quality Assurance (QA) Engineering (Python) A QA Engineer ensures the reliability, accuracy, and security of the pandemic modeling framework by developing and maintaining automated and manual testing processes. Their role is critical in validating epidemiological models, software components, and data integrity. https://www.pandemonium.dev/jobs/?id=PyQA
Human Mobility and Spatial Modeling Experts in this field analyze movement patterns and spatial data to simulate how diseases spread across different regions. Their work informs the development of dynamic coupling factors in our macro model. We are seeking a Human-Mobility Spatial Modeler: https://www.pandemonium.dev/jobs/?id=Mob
Public Health Policy Professionals in this area assess the implications of model outcomes on public health and guide policy recommendations. Their expertise ensures that the framework's outputs are relevant and actionable for decision-makers.
Behavioral Science/Psychology Experts in human behavior and psychology will guide the development of models that predict how people are likely to respond to public health policies, disease outbreaks, and other stressors. Understanding psychological drivers is key to modeling behavior accurately.
Computational Biology/Phylogenetics Experts in this field are essential for designing and implementing algorithms that can analyze genetic sequences and infer evolutionary relationships among various pathogens. Their expertise will be crucial in tracking pathogen evolution and potential transmission pathways.
Bioinformatics These specialists handle large datasets of biological information, especially genetic data. They will develop the tools and pipelines necessary to process and analyze sequence data efficiently.
Genomics Specialists in genomics will contribute to understanding the genetic factors that influence pathogen variability and resistance, crucial for predicting evolutionary trends.
Legal and Ethical Compliance Legal experts ensure that the framework adheres to data privacy laws and ethical standards, especially when handling sensitive health information. Their guidance is essential for maintaining user trust and legal compliance. We are seeking Legal Counsel: https://www.pandemonium.dev/jobs/?id=JD
Marketing and Public Relations Marketing professionals and community managers promote the framework, engage with stakeholders, and gather user feedback. Their efforts help in refining the tool and expanding its reach. We have openings for a Marketing Director and Social Media Manager. https://www.pandemonium.dev/jobs/?id=MktDir https://www.pandemonium.dev/jobs/?id=SMM https://www.pandemonium.dev/jobs/?id=Mkt
IT/DevOps/Platform Engineering T and DevOps professionals are responsible for managing the infrastructure that supports our modeling framework. They ensure continuous integration and deployment, system reliability, and scalability. Their expertise is crucial for automating workflows and maintaining the platform's performance.
Cybersecurity Cybersecurity experts protect our systems and data from potential threats. They implement security measures, conduct threat modeling, and ensure compliance with data protection regulations. Their role is vital in safeguarding sensitive health information and maintaining user trust.
Technical Support Engineering Technical Support Engineers provide essential assistance to framework users, including researchers and public health departments, ensuring they can effectively utilize the framework. Their responsibilities include troubleshooting issues, offering guidance on system functionalities, and ensuring a seamless user experience. Proficiency in Python is crucial for this role, as it enables the engineer to understand and resolve technical challenges related to the framework's codebase.
Administrative Support Administrative professionals assist with the coordination of research activities, manage communications, handle documentation, and support financial and compliance aspects. Their support enables the research team to focus on technical development and innovation.

Comments

No comments yet.

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 implications

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 / expertise required

biostatisticians
applied mathematicians
wastewater epidemiologists
environmental engineers with backgrounds in wastewater management

Comments

No comments yet.

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 implications

(No direct policy implications are immediately apparent)

Disciplines / expertise required

Sociologist
Behavioral Health
Medical Anthropologist
Social Network Modelling

Comments

No comments yet.

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 implications

If EWS exist, it allows the possibilities fo anticipation and preparation for outbreaks; even better, it might allow intervention to prevent outbreaks.

Disciplines / expertise required

understanding of disease outbreaks
data wrangler and analyst
some knowledge of EWS

Comments

No comments yet.

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 implications

improved predictions may be able to improve active management strategies for both public and private forests

Disciplines / expertise required

Forestry
Ecology
Entomology
Modeling
Economist Have you thought about including an economist who knows about the lumber industry? That might be important to potential management strategies.

Comments

Maybe you also need a statistician - while the proposed idea talks about causal models, it might be useful to analyze the data to discover subtle patterns that might have been missed before, especially if you can combine novel datasets or put old ones together in new ways...

N.H.F.1