Chain LLM Queries to Identify Required HealthKit Data

[PSchmiedmayer]

Problem

We currently package all the available HealthKit data to the LLM query for a user's request.

Solution

1. Query the GPT API to ask what health data would be needed to answer the question (a fixed set of types that the user gave permissions to and the needed date range)
2. Query the HealthKit API (with the selected type and date range)
3. Pass the requested data including the question to the GPT API

Code of Conduct

• I agree to follow this project's Code of Conduct and Contributing Guidelines

[PSchmiedmayer]

We should explore the functionality to query data added to the conversation in the context of StanfordSpezi/SpeziLLM#10 to include the function calling API of OpenAI.