From 'What' to 'How' and 'Why': Sharing LLM-Generated Retrospective Summaries of Older Adults' Passive Tracking Data with Remote Family Members

2026-06-02 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Human-Computer Interaction · Depth: Advanced, quick

Summary

A study explores using large language models (LLMs) to generate retrospective summaries of older adults' multi-modal passive tracking data for remote family members (RFMs). Addressing the challenge of synthesizing heterogeneous data, researchers customized the Vital Insight system to produce initial summaries across various dates and data availability scenarios. Following interviews with 11 RFMs, the system was redesigned into a multi-layer, multi-agent, insight-driven approach, evolving from objective statistics to enriched, context-aware narratives. A subsequent survey with the same 11 RFMs revealed significant improvements in satisfaction, perceived helpfulness, trust, and willingness to receive the redesigned summaries compared to initial versions. This work highlights the importance of shifting AI-generated summaries from merely presenting "What" data were collected to explaining "How" a loved one is doing and "Why".

Key takeaway

For AI Engineers developing systems for remote care monitoring, prioritize designing LLM-generated summaries that move beyond raw data presentation. Your systems should explain "How" and "Why" an older adult is doing, integrating contextual insights and personal knowledge. Iteratively gather feedback from family members to refine narrative generation, ensuring summaries are perceived as helpful, trustworthy, and actionable, thereby increasing their adoption and impact.

Key insights

LLM-generated summaries for remote family members require context-aware narratives beyond raw data to be helpful and trusted.

Principles

• Summaries must evolve from "What" to "How" and "Why".
• Incorporate personal knowledge for context-aware narratives.
• Multi-layer, multi-agent design enhances summary utility.

Method

Customized Vital Insight to generate initial LLM summaries. Redesigned into a multi-layer, multi-agent, insight-driven approach, building from objective statistics to context-aware narratives based on RFM feedback.

In practice

• Design LLM outputs to explain "How" and "Why", not just "What".
• Integrate stakeholder feedback for iterative system redesign.
• Structure summaries with objective data and contextual narratives.

Topics

• Large Language Models
• Multi-modal Data
• Remote Care Monitoring
• Human-Computer Interaction
• Retrospective Summarization
• Older Adult Care

Best for: AI Scientist, AI Engineer, Research Scientist

Related on AIssential

• RECOVER: Designing a Large Language Model-based Remote Patient Monitoring System for Postoperative Gastrointestinal Cancer Care — LLM-powered RPM systems can enhance postoperative GI cancer care through conversational agents and staff dashboards.
• RealICU: Do LLM Agents Understand Long-Context ICU Data? A Benchmark Beyond Behavior Imitation — RealICU is a new benchmark for LLMs in ICU settings, using hindsight-annotated data to overcome limitations of historical clinician actions.
• ChatHealthAI: Aligning Electronic Health Record Representations with Large Language Models for Grounded Clinical Reasoning — ChatHealthAI integrates EHR representations with LLMs for grounded, interpretable clinical reasoning and accurate patient prediction.
• Benchmarking GPT Models for Conversational AI Systems: Can AI Read a Doctor’s Notes? — LLMs struggle with medical abbreviation expansion, especially context-dependent ones, despite larger models performing better.
• Exploring the feasibility of conversational diagnostic AI in a real-world clinical study — Supervised conversational AI for pre-visit history taking is feasible, safe, and well-received by patients and clinicians.
• How I'm Using AI Today — A structured, principled approach to AI agent management mitigates LLM context limitations and enhances productivity.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.