Heart Disease Prediction API

Saving people, Saving lives

Comment 4

Heart Disease Prediction API

Heart Disease Prediction API is Machine Learning based solution to predict risks of heart disease using a integrable API.

Description

Our solution is based on microservices architecture and the AI Model can be independently used in any other platform with ease. We've also included a demo dashboard of our smart watch simulator where we can receive patient's real time health data directly from his smart watch and use those value to get a realtime prediction of immediate risks of heart attacks in a patient. This real time prediction is also simulated in a continuous graph to warn the patients in case of any risks.
In similar way, our api can be integrated with any smart watches, hospital management system, health apps to get a real time prediction of immediate risks of heart attacks.

Inspiration

Recently one of our friend’s sister died due to sudden cardiac arrest. And slowly we got to know that every family has lost at least one member due to the same threat.That incident totally struck into our nerves and we decided to do something in this specific domain but we were waiting for the right time. After registering for TO connect, we felt it’s the perfect time for us to start off with our vision and here we come up with this API.

What it does

So we have a CVD prediction dashboard.When the health data of patients are entered the API receives a post request and calls for the model. By analysing and comparing the values it predicts the percentage of heart disease risk and returns the values with confidence level. And the result is received in interpretable format.

How we built it

how we built it

Response

API Endpoint: http://18.215.165.176:8081/predict

{
  "status":"success",
  "value":"92",
  "risk":true,
}

status : If any error occured in backend it will say error else it will be set to success.
value : Confidence measure of the AI Model.
risk : boolean value which says if the risk of heart disease is high ( default threshold is 75% )

Challenges we ran into

Acknowledging the time frame we had to rush into everything which led us to mess up codes sometimes and staying whole night without sleep and working in low dopamine state was kinda tuff.

Accomplishments that we're proud of

We’ve finally built a successful product with high accuracy in a time frame of 24hrs. We can’t imagine where our product can reach if we put more effort in the coming days. We are absolutely a step closer to our vision.

What we learned

With consistency and a great team every toughest work becomes doable.

What's next for Heart Disease Prediction API

  • A voice based model to integrate in virtual assistant to monitor breathing while sleeping for a real time cardiac arrest prediction and warning system.
    • Optimize the AI by using all the available data in hospitals.
    • Predict and monitor additional lethal diseases.
Share this project:

Updates