AI in Healthcare

See how AI finds patterns in heart health.

Explore a real machine-learning model trained on public data. Change a fictional patient’s measurements and watch the prediction respond.

Try the predictor ↓ How does it learn?

✓ Real ML model✓ Explainable results✓ No data leaves this laptop

Live demonstration

Meet your sample patient

Choose a ready-made profile or adjust the measurements yourself. All profiles are fictional.

Patient information

Explain every field

Basics

Age years Sex in dataset

Measurements

Resting blood pressure mmHg Serum cholesterol mg/dL Maximum heart rate bpm ST depression oldpeak

Clinical test categoriesFeatures used by the public dataset⌄

Chest pain pattern Fasting blood sugar Resting ECG Exercise-induced angina Peak exercise ST slope Major vessels observed Historical thal scan result

▣ Runs locally · No patient data is stored

AI assessment

LOGISTIC REGRESSION

♥

Ready when you are

Choose a sample or enter measurements, then let the model find the closest learned pattern.

DATA→MODEL→RESULT

Behind the prediction

How does the AI learn?

The computer is shown examples, looks for mathematical patterns, then applies those patterns to a new sample.

▦STEP 01

Public data

270 anonymized examples from UCI’s Statlog Heart dataset.

→

⌁STEP 02

Train the model

Logistic regression learns how the 13 inputs relate to the known outcomes.

→

✦STEP 03

Find a pattern

A new fictional sample is compared with what the model learned.

→

◎STEP 04

Explain the result

The app shows the score and the inputs that influenced it most.

MODELLogistic regressionSimple, explainable classification

5-FOLD TEST ACCURACY84.4%Every record tested out-of-fold

ROC AUC89.9%Ability to separate the two classes

INTERNET NEEDEDNoneDesigned for an offline exhibition

SCIENCE GUIDE

What does every term mean—and how is it measured?

Explore all 13 inputs, decode a real dataset row, and learn the AI mathematics behind the prediction.

Open the detailed guide →

The important part

Powerful patterns need responsible people.

Real medical AI needs careful testing, diverse data, privacy protection, and professional oversight. A model can assist a healthcare professional; it cannot replace one.

AccurateExplainableFairSecureHuman-supervised