From Data to Impact: Convergence of Digital Technologies, AI/ML, and Human Factors for Population Health Solutions

Engineering advances in digital health require frameworks that link complex data streams, computational models, and human factors to measurable outcomes. This talk presents methods for moving from data to impact by embedding theory-driven constructs into high-dimensional analytics and plug-and-play low-latency Systems on Chip pipelines. We begin with multimodal sensing, electronic health records, wearables, and online communities, as pipelines for extracting behavioral and psychosocial signals. Next, we introduce AI/ML architectures, graph neural networks, topic modeling, and dynamic intent classifiers, that formalize mechanisms of engagement, diffusion, and risk prediction. These models enable scalable inference on communication patterns, adherence, and social influence, illustrated through diabetes self-management networks and peripartum depression discourse. Finally, we integrate human factors engineering and equity constraints into system design via frameworks such as Digilego, ensuring usability and fidelity across diverse populations. Multiple use cases demonstrating this translation from heterogeneous data into robust, adaptive interventions that reduce disparities and improve population health outcomes are discussed.