Clinical Prediction Models & The Multiverse of Madness

Are prediction models in healthcare reliable? Most models simply provide a single estimate of risk for an individual, but in this talk I raise awareness of the uncertainty of such predictions, framed as a 'multiverse' of possible models and predictions. I show that when sample sizes are small, the multiverse (uncertainty) might be vast, and so recommend this is routinely disseminated in practice, whenever a model is produced. This is important, as it helps researchers, guideline developers and patients to critically appraise existing models, and to understand the strength of evidence behind the predictions a model makes. I also discuss the pros and cons of providing uncertainty of risk estimates to patients, which some insight from real patient perspectives. Hope this is helpful! The works stems from recent papers including: Riley RD, Pate A, Dhiman P, et al. Clinical prediction models and the multiverse of madness. BMC Med 21, 502 (2023). doi: https://doi.org/10.1186/s12916-023-03... https://link.springer.com/article/10.... Riley RD, Collins GS, Kirton L, Snell KI, Ensor J, Whittle R, et al. Uncertainty of risk estimates from clinical prediction models: rationale, challenges, and approaches. BMJ 2025;388:e080749. https://www.bmj.com/content/388/bmj-2... Riley RD, Collins GS. Stability of clinical prediction models developed using statistical or machine learning methods. Biom J. 2023;65(8):e2200302. doi: 10.1002/bimj.202200302