SleepVariationsDetector

An initial phase to address any sleep anomalies from a wearable device more specifically from an iPhone running iOS operating systems.

Comment

Inspiration

Heard a lot of information about sleep disorders and sleep anomalies and how it can affect your life .But to actually study ones sleep variations you need the EEG scheduled . Using the actual device is a short study done in a week span but essential to cure the patient.What I thought was, if you want to address sleep disorders that is been happening for a long time and if the patient happens to have a wearable device and readings of your everyday sleep is collected with patient not conscious enough can help as an initial phase to study the issue.

What it does

User needs to provide the month and the year for which data is to be summarised. One month data is filtered and MCP server Calculates sleep efficiency , sleep percentages for each of the sleep stages(CORE,REM,AWAKE,DEEP),REM latency and Wake After Sleep Onset from apple health readings and leaves the data to be summarised by an AI model to provide an initial step towards getting your sleep variations at hand.

How we built it

-Prompt Opinion UI platform, -MCP Server developed using Python/Fast MCP -Integrated fast MCP server to PO Agent handler -Imported Apple Health Data corresponding to the patient on PO platform -Ngrok to expose the gateway API

Challenges we ran into

The error -LLM took too long to respond was one issue as a result of which multiple requests is sent by agent.Using MCP inspector to decipher what is happening helped resolve this issue . Small nuances like loading document references and understanding the scopes ,executing the prompt from the appropriate agents helped resolve issues to read uploaded documents and smooth execution of the intended task.

Accomplishments that we're proud of

Getting access to health data is not so easy task.The hackathon helped to understand the PO system and a MCP server to agent workflow

What we learned

Integration of MCP tool with an agent , aligning workflows and bridging the components to a platform like PO to accomplish results faster.

What's next for SleepVariationsDetector

Dowloaded Data to be cleaned up after a short cycle.More faster access to read the health data.Any other suggestions that could be projected from the existing day to day readings that has been collected. Currently only one month data retrieval is supported may need to enhance.

Built With

  • fastmcp
  • figma
  • promptopinion
  • python
Share this project:

Updates