PNNL scientists are tackling scientific challenges to improve human, animal, plant, and environmental health while combating the emergence of infectious diseases. Our experts excel at harmonizing complex data with artificial intelligence (AI) from across domains, providers, and modalities to create valuable and trustworthy data-driven insights for decision-makers.
Below are projects the PNNL One Health (OH) team is leading through key collaborations for situational awareness, early warning, and disease forecasting.
- Frameworks / Tenants
- Data Privacy / Standardization
- Multimodal Integration
- Situational Awareness
- Threat Forecasting
Frameworks / Tenants
Threat Risk and Event Analysis, Detection, and Surveillance framework (TREADS)
The Threat Risk and Event Analysis, Detection, and Surveillance framework (TREADS) offers an adaptable capability to meet new and changing threats compromising human, animal, plant, and environmental health, including:
- Situational awareness of current threats
- Early warning and risk assessment of unusual events
Prediction of upcoming threats and contributing factors.
Medical Intelligence (MedINT)
The MedINT (Medical Intelligence) platform integrates eight data sources on human, society, animal, and weather activities. The dashboard provides early warning and anomaly detection as well as a virtually assisted tool for deeper data investigation.
This flexible platform is based on data science, software engineering, and analytic visualization tools that enable detection, tracking, characterization, and situational assessment of events that have the potential to, or are currently causing, harm to human, animal, or plant health.
National Wildlife Disease Database (NWDD)
PNNL is working with the U.S. Geological Survey to develop a congressionally mandated National Wildlife Disease Database (NWDD) system to ultimately have better situational awareness and early warning for potential zoonotic disease threats from wildlife.
OH TREADS (One Health Threat Risk Event Analysis, Detection and Surveillance Framework)
PNNL’s OH TREADS (One Health Threat Risk Event Analysis, Detection and Surveillance Framework) tool is an adaptable capability to meet new and changing needs. Leveraging MedINT capabilities, OH TREADS integrates 16 additional data types on animal and environmental information and includes a forecasting risk map tool called Planner. OH TREADS combines different types of data and AI/machine learning (ML) techniques to provide situational awareness, threat prediction, and risk assessment to a variety of end users.
Data Privacy / Standardization
Trustworthy Rapid Approvals of Countermeasures (TRAC)
The Trustworthy Rapid Approvals of Countermeasures (TRAC) project is a multi-laboratory collaboration. PNNL is developing an AI data management system to extract, harmonize, and correlate across multimodal, cross-domain datasets. This Automated Quality Assurance and standardization will enable federated learning, which uses sensitive data without moving it from its source to preserve trust.
Policy Landscape Analysis Tool (PLAT)
Policy analysis tools help users understand complex requirements and jurisdictions (PLAT). For example, the web-based Biodefense Policy Landscape Analysis Tool (or B-PLAT) helps users explore and understand the U.S. biodefense enterprise through the lens of jurisdictional responsibilities and legal authorities. The platform maps more than 400 biodefense functions across 22 federal agencies and links them to relevant legal, policy, strategic, and public health authorities. In doing so, the PLAT serves as an integrated reference for understanding how legal authorities and agency responsibilities shape the national biodefense framework.
Multimodal Integration
Chem-bio Harmonization with Artificial Intelligence and Networks (CHAIN)
CHAIN (Chem-bio Harmonization with Artificial Intelligence and Networks) is a software tool that harmonizes complex and heterogeneous datasets. Using deep learning/ML and network methods, it enables users to explore data, discover complex relationships and patterns among data sources, and visualize data and model results with metrics of uncertainty and insights into model reasoning. Data supported by this application spans a variety of sensor types, including mass spectrometry-based ‘omics, Raman spectroscopy, ion mobility spectrometry, infrared spectroscopy, and photoionization detectors.
Situational Awareness
Biofeeds
Biofeeds is a biosurveillance platform designed to rapidly detect potential human, animal, and plant health threats by automatically harvesting over 70,000 multilingual global sources and using AI to identify relevant events and anomalies. It combines automated and analyst-driven labeling with standardized taxonomy and delivers timely reporting through alerts, subscriptions, and API access, currently supporting 183 users and 1,750 report recipients across 570 organizations.
Publication Literature Analysis and Text Interaction Platform for User Studies (PLATIPUS)
The Publication Literature Analysis and Text Interaction Platform for User Studies (PLATIPUS) was developed with funds from the COVID CARES ACT. The Department of Energy funded PNNL to build a public-facing, AI-driven platform that rapidly synthesizes rapidly growing publications to surface emerging insights with a One Health lens during fast-moving events, in this case COVID-19.
Street Wolf
The Street Wolf capability analyzes and categorizes scientific literature to identify new and emerging technologies relevant to mission needs. Using advanced analytics, it rapidly processes large volumes of research to surface key trends, innovations, and insights, enabling timely and informed decision-making.
Threat Forecasting
Machine Learning for Disease (ML4Dz)
For generalized disease forecasting, Machine Learning for Disease (ML4Dz) is a disease-forecasting system containing multiple AI/ML models. It generalizes global locations and various diseases through the integration of nearly 2,000 features across the One Health landscape.
VP Risk Maps
The VP Risk Maps capability transforms infectious disease forecasts into actionable risk maps to support operational awareness in complex environments. By integrating human, animal, and environmental data, it rapidly produces accurate and flexible outputs in minutes, significantly reducing analysis time while maintaining strong performance and adaptability.
Food & Agriculture Risk Models (FARM)
The Food & Agriculture Risk Models (FARM) provides advanced modeling capabilities to estimate agricultural crop damage and spread resulting from health threats, accounting for environmental conditions, geography, and mitigation strategies. It also assesses cyber vulnerabilities and consequences across food and agriculture systems, enabling scalable, data-driven risk analysis with uncertainty quantification to support decision-making.
AIMS-LEAF
AIMS-Leaf is utilizing AI/ML to integrate multiscale imaging and sequencing data for predictive modeling of host reactions, in this case for plant heat stress tolerance. The approach aims to build flexible, scalable AI frameworks that link genetic modifications to phenotypic expression, for a faster and clearer picture of host responses.
Additional AI Capabilities
PNNL has also developed additional capabilities as stand-alone applications for other purposes that are now being incorporated into the TREADS toolbox to enhance our ability to achieve early warning of One Health threats.