JEEVHITAA -- An End-to-End HCAI System to Support Collective Care

2025-12-06

AI Summary

Current mobile health platforms are individually centered, hindering multi-stakeholder collaborative and auditable collective care decision-making. To address this, we propose a human-centered AI system for collective care. Methodologically, we (1) design a role-aware, time-constrained, fine-grained access control mechanism enabling relationship-sensitive information sharing; (2) establish a verifiable information flow framework coupled with evidence-based credibility assessment for health content; and (3) develop an end-to-end prototype integrating Android/Flutter, Google Health Connect, BLE sensors, on-device end-to-end encrypted storage, and Firebase synchronization—enhanced by a hierarchical registration and retrieval-augmented LLM pipeline. The system is validated via ontology-driven simulation and synthetic data, demonstrating robustness, interoperability, and long-term deployability. It establishes a novel paradigm for collaborative health decision-making in home and community settings.

Technology Category

• Humans and AI: Planning and Decision Support for Human-Machine Teams
• Machine Learning: Hardware-aware ML
• Multiagent Systems: Mechanism Design

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems

Abstract

Current mobile health platforms are predominantly individual-centric and lack the necessary primitives for coordinated, auditable, multi-actor workflows. However, in many settings worldwide, health decisions are enacted by multi-actor care networks rather than single users. We present JEEVHITAA, an Android/Flutter system that provides context-sensitive, role-aware sharing and verifiable information flows for care circles. JEEVHITAA ingests platform and device data (via Google Health Connect and BLE connectors), constructs multi-layer user profiles from sensor streams and tiered onboarding, and enforces fine-grained, time-bounded access control across permissioned care graphs. Data are end-to-end encrypted in local stores and during peer sync (Firebase), and provisions are made for document capture by camera or upload as PDF. An integrated retrieval-augmented LLM pipeline (i) produces structured, role-targeted summaries and action plans, (ii) enables users to gather advanced insights on health reports, and (iii) performs evidence-grounded user-relevant verification of arbitrary health content, returning provenance, confidence scores, and source citations. We describe the system architecture, connector abstractions, and security primitives, and evaluate robustness and compatibility using synthetic, ontology-driven simulations and vendor compatibility tests. Finally, we outline plans for longitudinal in-the-wild deployments to measure system performance, the correctness of access control, and the real-world effectiveness of relationship-aware credibility support.

Problem

• Supports coordinated multi-actor healthcare workflows
• Provides verifiable, role-aware data sharing in care circles
• Enables evidence-based verification and insights for health content

Innovation

• End-to-end encrypted multi-actor care workflows
• Fine-grained time-bounded access control graphs
• Retrieval-augmented LLM for role-targeted summaries