AI Health Management

Problem

Health reports are often stored as PDF, images, or Word files, and abnormal indicators are usually presented as a single reference range rather than an actionable plan. Users still need to interpret results, assemble exercise suggestions, find instructional videos, log daily check-ins, and compare trends across multiple reports.

How the skill works

The skill is triggered by natural language, such as "analyze my health report", "my blood pressure is high", "generate an exercise plan", or "log today's workout". It reads user context, extracts indicators such as ALT, GGT, lipids, blood glucose, and blood pressure, and masks sensitive fields such as name, ID number, and phone number. It then flags abnormal values against reference ranges, for example associating elevated ALT/GGT with fatty liver tendency, and elevated lipids, glucose, or blood pressure with corresponding risk areas.
Based on these signals, it generates exercise plans, such as Baduanjin, tai chi, brisk walking, cycling, swimming, shoulder and neck stretching, or strength training, and links matching instructional videos from Bilibili. Operationally, it can deliver daily plans and reminders through OpenClaw Heartbeat or cron, record check-in data, compare indicator trends across reports, and update the exercise plan when meaningful changes are detected. Weekly or monthly summaries and Feishu document output are part of the check-in review workflow.

Boundaries

It is intended for personal health tracking, report organization, and exercise reminders, and it does not replace medical diagnosis. Accuracy for scans, blurry screenshots, or complex layouts depends on OCR quality. The current version is a v1.0 test release, and some reminders, file formats, and exercise outcome evaluation remain pending.