从苹果健康数据构建一个AI健康教练
Build an AI health coach from Apple Health data
作者将苹果健康数据（营养、活动、体重秤、血压、睡眠、锻炼）导入自己的服务器/代理，以充当健康教练.
做了什么
The poster feeds Gemini a weekly manual upload of their personal health data into an ongoing chat that already contains background on them: food logs, steps and activity, smart scale and blood-pressure readings, sleep, workout logs, and supplements. They review the data before uploading to catch bad entries, then use the model for specific recommendations like warm-ups, workouts, cool-downs, joint-specific advice, and a nutrition plan built around foods they actually keep on hand.
效果如何
They report concrete wins: Gemini flagged very low iodine intake, they switched to iodized salt and felt better within a week, and it reorganized five weekly workouts into a PPL, Tabata, and distance-run split that made recovery easier. Limits noted in the thread: the setup is manual, and long-running chats can start forgetting earlier weeks.
值得注意
The main catch is upkeep: they only upload once a week on purpose so they can manually sanity-check entries before sending them.
试试这个 Prompt
Help me build a personal health coach that reads my exported Apple Health data (nutrition, activity, weight, blood pressure, sleep, workouts) from [export file or folder path]. Write a script that parses it and produces a weekly summary with trends and three suggestions for my goal of [goal]. You're done when I can run one command and get a readable report, and don't send my data to any outside service without showing me first.