Back to Selected Work
TECHNICAL CASE STUDY
Heart Disease Prediction System
Clinical Machine Learning Classification & Diagnostic Dashboard
Executive Overview
An interactive machine learning application that predicts patient heart disease risk based on key clinical parameters such as cholesterol, resting blood pressure, max heart rate, and ST depression.
The Problem
Early clinical risk screening often requires parsing multiple diagnostic data points. An automated, accessible ML screening dashboard helps highlight risk probabilities rapidly.
My Engineering Role
ML Developer — Cleaned clinical datasets, trained and evaluated classification models, engineered features, and built the interactive Streamlit dashboard.
System Architecture Flow
[01]Patient Clinical Inputs → Preprocessing & Feature Scaler (Scikit-Learn)
[02]Trained ML Classification Model (Random Forest / Logistic Regression) → Probability Prediction
[03]Streamlit Dashboard → Interactive Visualization & Risk Metrics Display
Key Architectural Decisions
Streamlit for UI
Allowed rapid deployment of an interactive Python-native dashboard directly connected to the scikit-learn model pipeline.
Verified Outcomes & Results
- ✓Deployed a live, accessible prediction tool at heardiseaseprediction1811.streamlit.app.
- ✓Achieved high accuracy on test clinical data split.
Lessons Learned
- •Medical ML applications must prioritize recall to avoid missing high-risk cases.
- •Streamlit is exceptionally effective for prototyping ML tools quickly for user feedback.