Prediction study overview — Clinical prediction model development

Intended use

Define the clinical decision, target population, prediction time and outcome horizon; assess existing models first.

Protocol

Prespecify candidate procedures, sample-size rationale, validation and evaluation before fitting models.

Define data and prediction time

Choose suitable sources and align eligibility, predictor availability and outcome ascertainment with the prediction time.

Prepare predictors and outcomes

Define transformations and missing-input handling that can also operate at deployment; prevent future-information leakage.

Develop model

Fit the planned model pipeline with controlled complexity and tuning separated from evaluation.

Internally validate

Repeat the entire development process in resampling, respecting clustering and chronology, to estimate optimism.

Assess performance

Evaluate calibration, discrimination, overall error and clinical usefulness with uncertainty at the intended horizon.

Report

Report the complete model and preprocessing, validation and limitations using TRIPOD+AI; plan independent external evaluation.

Prediction study overview

End flow

Protocol and decision register

Record the protocol, approvals and registration where applicable. Preserve planned decisions and amendments separately from results.

Protocol and rationale evidence

Attach the dated protocol, model or test specification, approvals and amendments. Identify any applicable study registration, registry and identifier explicitly. Diagnostic accuracy and prediction studies are not automatically interventional clinical trials; a disease registry is a data source, not protocol registration.