Method

Model Training

Manage datasets and configure benchmarking parameters for clinical classification.

Platform Note: This demonstration environment uses an ephemeral filesystem. Newly uploaded datasets and trained models will be reset upon session timeout or application restart.

1. Data Management

Upload new clinical datasets and define metadata for the study.

Available Datasets

Label	Source
Cleveland Heart Disease Cleaned Dataset	Cleveland Clinic Foundation (Processed)

2. Model Training

Select a verified dataset and define the learning architecture.

Training Input Guide

Stratified cross-validation is shared across all methods. KNN also uses the neighbor count; the other methods keep their defaults fixed in this UI.

K-Nearest Neighbors

Inputs: dataset, validation folds, random state, and K neighbors.

Meaning: the model compares each case to the nearest training examples and predicts by vote.

Naive Bayes

Inputs: dataset, validation folds, and random state.

Meaning: the model scores each class probabilistically from the feature values.

Support Vector Machine

Inputs: dataset, validation folds, and random state.

Meaning: the model learns a separating boundary between the classes.