Portfolio / Ejection fraction predictor

Ejection fraction predictor

A machine learning model that predicts cardiac ejection fraction with 97% accuracy and explains each prediction.

Preview of Ejection fraction predictor
97%prediction accuracy
1,600+patient records
SHAPexplanation for every prediction

The problem

Clinicians need predictions they can trust, which means high accuracy and a clear reason behind every result.

What I built

  • A full pipeline from raw patient data to a deployed model
  • Feature engineering and model selection on 1,600+ patient records
  • SHAP explanations showing which factors drove each prediction
  • An interactive demo app

The result

An accurate model whose predictions clinicians can read and question.

Have a project in mind?

Email talha@pokitbio.com. I reply within one working day.

Start a project