AI-Driven Digital Health Behaviour Change at Nation Scale

Mission

---

This workshop addresses a critical gap identified at previous Digital Public Health conferences: while pilot studies demonstrate technical feasibility of AI-driven interventions, progression from successful pilots to population-scale implementation remains rare and slow. Most digital health platforms employ deterministic, rule-based algorithms that achieve initial adoption but fail to maintain long-term adherence – with over 75% of health applications demonstrating sustained adherence rates below 30%.

This workshop examines the transition to personalization architectures proven at national scale, contributing peer-reviewed methodology and econometric validation frameworks that address temporal misalignment between prevention investments and benefit realization within healthcare financing constraints. The goal for the workshop is to advance implementation science by examining deployments across different healthcare systems – government-led, public health infrastructure, and integrated payer-provider models – demonstrating replicability patterns essential for translation from pilot to population scale.

This session directly addresses some of the priorities identified in prior DPH research: bridging the gap between AI capability and validated deployment; establishing evidence standards beyond pilot metrics; addressing health equity through demonstrated outcomes across diverse populations; and providing actionable frameworks for health system executives, payer organizations, and government health agencies implementing value-based care programs requiring outcome attribution and budget justification.

Objectives

---

Digital health and AI advancement stand at critical crossroads. While AI capabilities have advanced exponentially, most digital population health interventions continue to rely on deterministic, rule-based algorithms that have consistently failed to achieve sustained engagement or population-scale impact. These if-then systems, which segment populations by demographics rather than learning individual behavioral patterns, cannot adapt to personal context, intelligently prioritize across competing health needs, or orchestrate interventions across multiple chronic conditions.

This gap between AI potential and deployment reality manifests in three persistent barriers: sustained engagement remains elusive despite proliferation of connected health devices; prevention strategies face structural economic misalignment within systemic incentives and annual budget cycles; and deployments remain trapped between technology fragmentation and unsustainable human resource models.

The Adherence Gap.

Sustained patient engagement remains the primary barrier to behavior change interventions. Despite proliferation of connected health devices providing continuous biometric data—continuous glucose monitors, blood pressure monitors, activity trackers—most patients fail to maintain long-term adherence to therapeutic regimens or lifestyle modifications. Current digital health platforms predominantly employ deterministic, rule-based intervention systems that segment populations by demographic characteristics rather than individualizing based on behavioral phenotypes, contextual factors, and personal preferences.

The Economic Misalignment.

Prevention strategies face structural economic barriers within healthcare financing systems. Annual budgeting cycles and high member turnover rates create temporal misalignment between intervention costs and benefit realization, making long-horizon prevention investments economically irrational when beneficiaries may transition to competing payers.

The Deployment-Scale Paradox.

Digital health implementations face a persistent technology-versus-human resource dilemma. Technology-centric approaches generate application fragmentation with limited clinical workflow integration. Human-intensive models employing care coordinators and health coaches achieve personalization but cannot scale economically. Existing algorithmic approaches lack the adaptive intelligence required to orchestrate interventions across multiple comorbidities or dynamically adjust to individual contexts.

These challenges share a common limitation: deterministic intervention algorithms cannot learn individual behavioral patterns, incorporate personal context, or intelligently prioritize among competing clinical needs. Recent advances in Graph Neural Networks, knowledge representation, and cloud-based infrastructure create unprecedented opportunity to address these persistent barriers.

This panel brings together government health agencies implementing nation-scale adaptive health systems, public health system executives integrating personalized interventions into clinical infrastructure, integrated health system innovators orchestrating ecosystem platforms across multiple programs, technology partners enabling cloud-scale deployment while maintaining data sovereignty, and clinical leadership establishing evidence standards for AI validation.

Intended Audience

---

This workshop is designed for stakeholders navigating population-scale deployments:

• Health System Executives and Clinical Leaders: Seeking validated frameworks for evaluating AI-driven engagement platforms and understanding integration requirements within clinical workflows.

• Payer Organizations: Including Medicare Advantage, Medicaid, and commercial insurers requiring economic validation methodologies and sustainable financing models for value-based care.

• Government Health Officials and Policymakers: Implementing or considering national-scale health initiatives and examining proven deployment patterns across different regulatory environments.

• Healthcare IT and Innovation Leaders: Evaluating technical architectures for adaptive systems, understanding cloud-scale infrastructure requirements, and assessing data sovereignty frameworks.

• Digital Health Researchers and Data Scientists: Contributing peer-reviewed methodologies, examining Graph Neural Networks and knowledge representation approaches, and advancing implementation science frameworks.

• Investment Community, Influencers and Strategic Advisors: Assessing market validation signals, understanding sustainable business models, and evaluating evidence quality distinguishing validated deployments from pilots.

The expected participants are drawn from across the broad DPH audience, including researchers, frontline health and care staff, commissioners, and technology experts.

Expected outcomes

---

Expected Takeaways for participants:

• Technical frameworks for evaluating LLM based Health
  Coaches, Graph Neural Networks and knowledge representation approaches that enable individual-level personalization at population scale, distinguishing adaptive architectures from deterministic segmentation systems.

• Economic validation methodologies for demonstrating return on investment within annual budgeting cycles, including outcome measurement and attribution approaches that address temporal misalignment between prevention investments and benefit realization.

• Deployment patterns proven across government health agencies, public health systems, and integrated health organizations that resolve the technology-human resource paradox through ecosystem platforms balancing automation and clinical oversight.

• Partnership structures that align divergent stakeholder incentives across government, health systems, payers, and technology providers while addressing data sovereignty requirements and enabling sustainable business models.

• Evidence standards for distinguishing validated, longitudinal deployments from pilot initiatives, including criteria for peer-reviewed methodology, population-scale operation, and replicability across diverse healthcare contexts.

Format

---

The workshop format – combining paper presentations, expert panel dialogue, and poster discussions – will encourage active engagement and facilitate questions and comments throughout the session.

Speakers

---

---

---

---

---