PROBAST+AI

A structured tool for judging the quality, risk of bias and applicability of studies that develop or test a clinical prediction model, including AI models.

What it is¶

PROBAST+AI walks you through signalling questions in four domains: participants and data sources, predictors, outcome, and analysis. It has two parts: one rates the quality of model development, the other rates the risk of bias of a model evaluation, and both rate whether the study applies to your own question. It updates PROBAST (2019) and covers both regression and AI methods. It was built mainly for systematic reviews, but you can also use it to check your own study design or to appraise a model before you adopt it.

In healthcare¶

Made for diagnostic and prognostic models in clinical care; a useful check before a hospital takes up a published model or a vendor’s validation evidence.

Where to use it in The ELSA Way¶

• External and multi-site validation › Types of external validation: Its questions on participants, setting and outcome help you judge whether a validation study really tests the model on a new population or only looks like it does.

• Evaluation planning › Pre-specification: Running the questions on your own planned study shows the design and analysis choices that would raise the risk of bias, while you can still change them.

References¶

• Moons KGM, Damen JAA, Kaul T, Hooft L, Andaur Navarro C, Dhiman P, et al. PROBAST+AI: an updated quality, risk of bias, and applicability assessment tool for prediction models using regression or artificial intelligence methods. BMJ. 2025;388:e082505. Moons et al. (2025)

Related tools¶

• TRIPOD+AI: A 27-item checklist for writing up studies that develop or validate a clinical prediction model, whether it uses regression or machine learning.