Introducing MAIA - Medical AI Assistant on Internet Computer

Hi everyone!

We’re excited to introduce MAIA, the Medical AI Assistant on Internet Computer.

  • Our Story

    We started with a simple goal: to provide patients with valuable insights and ensure the accuracy of their diagnoses, offering a reliable “second opinion.” But as we continued to develop, we realized that we could do so much more.

    We envisioned a tool that could not only help patients but also assist doctors in making their diagnostic processes more efficient and accurate. This led us to start building MAIA.

  • What Can MAIA Do? [Summary of First Sprint]

    • Brain CT Scans: Upload a Brain CT Scan, and MAIA can analyze it to detect glioma, meningioma, and pituitary tumors or confirm if the scan is clear.

    • Kidney CT Scans: Upload a Kidney CT Scan, and MAIA can identify cysts, stones, or tumors, or confirm if the scan is normal.

    • Lung CT Scans: Upload a Lung CT Scan, and MAIA can check for adenocarcinoma, large cell carcinoma, and squamous cell carcinoma, or verify if the lungs are healthy.

    • Skin Lesion Images: Upload an image of a mole, and MAIA can determine if it’s benign or malignant, providing quick and reliable feedback.

    • Chest X-Ray: Upload a Chest X-Ray, and MAIA can diagnose tuberculosis, coronavirus, bacterial pneumonia, viral pneumonia, or confirm if the lungs are healthy.

    • Provide Diagnosis Information: Get detailed information about the diagnosis provided by MAIA.

    • Answer Questions: Ask questions about your diagnosis, and MAIA will provide informative answers.

  • Planned Features for the Next Sprint

    • Ultrasound Scan Diagnoses: MAIA will assist in analyzing and classifying ultrasound scans to detect issues like breast cancer and lymph node problems.

    • Comprehensive Skin Analysis: Right now, MAIA can analyze moles to detect benign or malignant growths. We want MAIA to be able to analyze all skin-related diseases and problems.

    • Bone Health Assessments: Including the detection of osteoporosis and bone fractures.

    • Blood Test Analysis: MAIA will help interpret blood test results to diagnose various conditions.

    • UI Updates: We want MAIA to have an easy-to-use interface for mobile users. Therefore, we will be making updates to improve the user experience.

  • Demo

    MAIA - Medical AI Assistant on Internet Computer - Watch Video

Our journey with MAIA is just starting. We’ll continue to share updates about MAIA here.

We are open to feedback and suggestions during these early stages, and we would love to hear your thoughts.

Be one of the early supporters and follow MAIA on X here.

6 Likes

I like your idea for several reasons. First is that it’s not defi. I think ICP needs to diversify in terms of the types of projects it has. It’s too defi focused although I understand that’s where a lot of the money is at for crypto.

Another reason I like your project is because I can see Ai being integrated more into the medical field. I read a study where it was determined that doctors were copy/pasting from ChatGPT in their articles for academic journals. Surely, they wouldn’t mind also using Ai to aid in diagnoses, as well. In fact, some of them do rely on WebMD as a second opinion, but instead of Googling for the information themselves, an Ai project like yours, can sort of automate that process by retrieving the information for them.

I hope you introduce yourselves more, your background, and talk more on the need for this project, it’s potential in the market, etc. LFG @ICP.

5 Likes

This is impressive. I look forward to seeing your next milestone. Thanks for sharing.

4 Likes

Thank you so much for your interest and valuable feedback!
We’ll also be sharing the details you requested with our second sprint update. Stay tuned :rocket:

Thank you!
We’ll be sharing more updates with forum soon. We appreciate your support.

1 Like

Hi, nice work!
I’m curious on the tech side of things, are you storing the model on chain?

1 Like

Hi! Yes, we are storing our diagnose models on chain.
Thank you for your interest!

1 Like

All the best with the project!

1 Like

Good project definitely helpful
Just curious what models are you using for diagnosis are they opensource ?
Are you planning to fine tune the models furthermore based on users data ?

1 Like

Hi Txo, Thank you for your interest !

We are currently training our models from scratch. However, we are considering open-sourcing them in the future.
While our current plans do not include storing user data for fine-tuning purposes, we do intend to create a feedback system to identify models that need further fine-tuning and improvement.

You can review our detailed plans in the grant application we recently submitted. We look forward to your feedback!