Precision Medicine and Member Care Optimized by Causal AI
Causal Inference
- Moving beyond predictive analytics to understand the cause-and-effect relationship in treatment pathways.
Time-Series Expertise
- Specialized architectures for high-frequency ICU data and longitudinal member claims.
Privacy-First AI
- Ensuring all models comply with HIPAA and clinical data integrity standards.
Hybrid Digital Twin Synthesis
- Creating virtual cohorts to simulate clinical trial outcomes and reduce R&D and clinical trial costs.
Real-World Results
See how we've delivered measurable impact
Mcure Therapeutics
Results
Working with Mcure clinical and data scientists to deploy digital twins in an upcoming clinical trial.
Challenge
Understanding the impact of dynamic, high-stakes treatment pathways in intensive care environments where real-world evidence is often fragmented.
Solution
We developed a hybrid digital twin of the ICU patient environment. By utilizing multi-headed transformers and time-series clustering, the platform captures the nuances of physiological changes over time. Causal AI identifies the direct impact of specific treatment interventions on patient outcomes. Using time-series clustering of transformer outputs and hidden Markov models, we accommodate heterogeneities in patients and pathogenesis. We combine biological and data-driven modeling approaches to simulate clinical pathways and impacts of counterfactual treatments.
Humana
Results
Deployed in operations and won the best-of-breed technology award from ComputerWorld magazine.
Challenge
Shifting from reactive care to proactive, integrated health management for millions of members.
Solution
Deployed an Integrated Health Management Platform designed to synchronize disparate member data into a unified care strategy. We applied sophisticated time-series feature engineering to capture longitudinal health trends, fed into deep learning networks to identify high-risk transitions. By applying causal AI models, the platform moves beyond correlation, allowing Humana to determine which specific care interventions yield the best long-term health improvements for individual members.
PeaceHealth & NIH
Results
Participants in a randomized controlled trial achieved statistically significant weight loss while improving physical activities from 3K to 8.5K steps per day. The research team published over 40 research papers in peer-reviewed journals.
Challenge
Leveraging social networks to spread healthy behaviors and improve community health outcomes under a 3-year NIH-funded study.
Solution
In collaboration with university partners and PeaceHealth, we deployed a pioneering Personal Health Social Network integrated with wearable sensors to measure physical activities and weight. We utilized deep learning networks to analyze sensor data and social interactions in real-time. By deploying causal AI models, we mapped how healthy behaviors spread through a social graph, allowing for targeted interventions that utilize social influence and gaming to improve community-wide wellness metrics.