HealthGPT

HealthGPT is an experimental iOS app based on Stanford Spezi that allows users to interact with their health data stored in the Apple Health app using natural language. The application offers an easy-to-extend solution for those looking to make large language model (LLM) powered apps within the Apple Health ecosystem.

HealthGPT is an open-source project of the Stanford Biodesign Digital Health team. The initial prototype based on Spezi and the SpeziTemplateApplication was built by Varun Shenoy.

[!NOTE] Do you want to try HealthGPT? You can download it to your iOS device using TestFlight!

Features

• Extensible architecture built on the Stanford Spezi open-source digital health development framework for easy customization.
• Chat-style interface for user-friendly health data interaction using the SpeziChat module with speech-to-text (recognition) as well as text-to-speech (synthesize) accessibility capabilities and chat export functionality.
• Integration with the Apple Health app via SpeziHealthKit.
• GPT-3.5 + GPT-4 queries through the SpeziLLM module.
• Option for privacy preserving local execution of LLM queries using SpeziLLMLocal with Llama3 8B, including automated download and storage of model files during onboarding.
• Out of the box support for querying sleep, step count, active energy, exercise minutes, heart rate, and body mass.

Disclaimer

HealthGPT is provided for general informational purposes only and is not intended as a substitute for professional medical advice, diagnosis, or treatment. Large language models, such as those provided by OpenAI, are known to hallucinate and at times return false information. The use of HealthGPT is at your own risk. Always consult a qualified healthcare provider for personalized advice regarding your health and well-being. Aggregated HealthKit data for the past 14 days will be uploaded to OpenAI. Please refer to the OpenAI privacy policy for more information.

Set Up

Building and running HealthGPT requires a Mac with Xcode 16.2 or newer installed.

1. Clone this repository to your local computer.
2. Open HealthGPT.xcodeproj in Xcode. Wait for all dependencies to install and indexing to finish.
3. Run the app (on an iOS device or in the iOS simulator) and play with HealthGPT on your own data 🚀

[!IMPORTANT]
If you are running HealthGPT using the simulator, you will need to manually add data in the Apple Health app. Otherwise, all of your results will read zero.

[!IMPORTANT]
Running HealthGPT with a local LLM requires installing the app on a physical device, as SpeziLLMLocal requires a modern Metal MTLGPUFamily and simulators do not provide that.

You can add queries for additional HealthKit quantities and categories as follows:

1. Update the SpeziHealthKit configuration in HealthGPT/HealthGPTAppDelegate.swift to include the additional data type(s). For more information about configuring SpeziHealthKit, please refer to the official documentation.
2. Edit HealthGPT/HealthGPT/HealthDataFetcher.swift to create appropriate query for your data type(s).
3. Update the prompt in HealthGPT/HealthGPT/PromptGenerator.swift to pass the newly acquired data to the OpenAI API.

Network-local privacy-focused Fog LLM Mode

HealthGPT can run LLM inference on nearby machines in your local network, called fog nodes, instead of only on-device or in the cloud. The app discovers a fog node via mDNS, connects to it, and streams model responses back while dispatching inference tasks. This setup provides low latency, strong performance, and improved privacy, since models run on your own hardware inside your own network.

For instructions on running a minimal Docker-based fog node on Linux or macOS, see FogNode/README.md.

Contributing

Contributions to this project are welcome. Please make sure to read the contribution guidelines and the contributor covenant code of conduct first. You can find a list of contributors in the CONTRIBUTORS.md file.

License

This project is licensed under the MIT License. See Licenses for more information.