DEV Community

Beck_Moulton
Beck_Moulton

Posted on

Claude-Ally-Health: Your Private AI Medical Assistant & Health Data Manager

#ai #python #mentalhealth

In an era where personal health data is often scattered across physical paper reports, various hospital apps, and mental notes, keeping a cohesive medical history is a challenge. Enter Claude-Ally-Health, an open-source project that leverages the power of Claude AI to create a private, intelligent, and file-based Personal Health Information System (PHIS).

In this post, we'll explore what makes this tool unique, how it works, and why it might be the perfect solution for privacy-conscious developers looking to manage their health data.

πŸ₯ What is Claude-Ally-Health?

Claude-Ally-Health is a project developed by WellAlly Tech. It serves as an intelligent healthcare assistant that combines Claude AI technology with medical expertise.

Unlike traditional health apps that store your sensitive data in the cloud, Claude-Ally-Health operates on a pure file system (JSON) approach. It runs locally using the Claude Code command-line interface, ensuring that your data remains private and in your control.

Key Philosophy

  • Privacy First: No external databases. All data is stored in local JSON files.
  • AI-Powered: Uses computer vision and LLMs to read medical reports and offer consultations.
  • Structured Data: automatically converts unstructured text and images into structured health records.

✨ Core Features

The system is packed with features that rival commercial electronic medical record (EMR) systems, but tailored for personal use:

1. πŸ“‚ No Database Required

Forget about setting up PostgreSQL or MongoDB. The system uses a clean directory structure with JSON files to store profiles, medication history, and lab results. This makes backing up your health data as simple as zipping a folder.

2. πŸ–ΌοΈ AI Report Recognition

Received a lab result or an imaging report? simply point the tool to the image. It supports:

  • OCR & Extraction: Automatically pulls biochemical indicators and reference ranges.
  • Imaging Structure: Extracts structured data from CT/MRI reports.
  • Discharge Summaries: Digitizes and stores hospital discharge notes.

3. πŸ‘¨β€βš•οΈ Multidisciplinary Team (MDT) Consultation

This is perhaps the coolest feature. The system simulates a hospital's Multidisciplinary Team. It includes 9 specialist AI agents (Cardiology, Neurology, Oncology, etc.) that can "discuss" your case to provide a comprehensive analysis.

4. πŸ’Š Drug Safety & Interaction Checks

The system includes a built-in safety module that checks for:

  • Drug-Drug Interactions: Flags potential risks between medications.
  • Severity Levels: Uses a 5-level warning system (A/B/C/D/X).
  • Allergy Management: Checks new meds against your recorded allergies.

5. ☒️ Radiation Dose Tracking

A rare feature in personal health apps: it tracks your cumulative radiation exposure from medical imaging (X-rays, CTs) using body surface area adjustments and decay models.

πŸ› οΈ How It Works

The system relies on Claude Code Slash Commands. You interact with it via the terminal using natural language or specific commands.

Directory Structure

Your health repository looks like this:

my-his/
β”œβ”€β”€ data/
β”‚   β”œβ”€β”€ profile.json           # Basic user info
β”‚   β”œβ”€β”€ medications/           # Drug records
β”‚   β”œβ”€β”€ radiation-records.json # Radiation tracking
β”‚   └── ...
└── .claude/
    └── commands/              # The magic logic files
Enter fullscreen mode Exit fullscreen mode

Usage Examples

1. Setting up your profile:

/profile set 175 70 1990-01-01
# Sets height (cm), weight (kg), and DOB
Enter fullscreen mode Exit fullscreen mode

2. Saving a medical report:
Don't type in numbers manually. Just feed it the image.

/save-report /path/to/blood_test.jpg
Enter fullscreen mode Exit fullscreen mode

The AI analyzes the image, creates a JSON entry in data/η”ŸεŒ–ζ£€ζŸ₯/, and indexes it.

3. Checking Drug Interactions:

/interaction check
# Automatically checks your current medication list for conflicts
Enter fullscreen mode Exit fullscreen mode

4. Starting a Consultation:
If you have complex symptoms, you can invoke the specialist system:

/consult
# Starts the Multi-Disciplinary Team analysis
Enter fullscreen mode Exit fullscreen mode

πŸš€ Getting Started

Since this is a Python-based tool utilizing Claude, you need to have Claude Code installed.

  1. Clone the Repository:

    git clone https://github.com/huifer/Claude-Ally-Health.git
cd Claude-Ally-Health

  2. Open Claude Code:
    Run the tool in the current directory (ensure you have the necessary API access/environment set up for Claude).

  3. Initialize:
    Start by setting your basic physical parameters using the /profile command.

⚠️ Important Safety Note

While this tool is impressive, the developers include a critical disclaimer that everyone should heed:

  • Not a Doctor: This system does not replace professional medical advice.
  • No Prescriptions: It does not prescribe medication doses.
  • Reference Only: All analysis is for personal reference and health management support.

πŸ’‘ Conclusion

Claude-Ally-Health is a fascinating example of how AI Agents and LLMs can be applied to practical, high-value personal workflows. By keeping data local and structured, it empowers users to take ownership of their medical history without sacrificing privacy.

Whether you are a developer looking to track your own health metrics or an enthusiast exploring AI agents, this project is definitely worth a star.

πŸ”— Links:

Top comments (0)