Consensus statement on the credibility assessment of machine learning predictors - Position Article

Aug 4, 2026

Upload by Levi

Rapid integration of machine learning (ML) predictors into in silico medicine is reshaping estimation of quantities of interest that are difficult to measure directly. Credibility becomes essential when predictions support high-stakes healthcare decisions. This position paper summarizes an expert consensus from the In Silico World Community of Practice, presenting 12 key statements for evaluating credibility, stressing causal knowledge, rigorous error quantification, robustness to biases, and guidance for researchers, developers, and regulators.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages7

Reading Time18 min

AccessFree to download

Introduction

Background and need for credibility assessment

Consensus development

Expert process and scope

Briefings in Bioinformatics, 2025, 26(2), bbaf100 [https://doi.org/10.1093/bib/bbaf100](https://doi.org/10.1093/bib/bbaf100) Position Article Consensus statement on the credibility assessment of machine learning predictors Alessandra Aldieri1 ,‡, Thiranja Prasad Babarenda Gamage2 , Antonino Amedeo La Mattina3 ,4 , Axel Loewe5 , Francesco Pappalardo 6 , *, Marco Viceconti7 1 Department of Mechanical and Aerospace Engineering, Politecnico di Torino, Corso Duca degli Abruzzi, 24-10129 Torino, Italy 2Auckland Bioengineering Institute, University of Auckland, Private Bag 92019, Auckland 1142-New Zealand 3 Medical Technology Laboratory, IRCCS Istituto Ortopedico Rizzoli, Via di Barbiano, 1/10-40136 Bologna, Italy 4Yi Li National Clinical Research Center for Aging and Medicine, Huashan Hospital, Fudan University, 512 Huashan Rd, Jing’An, 200031, Shanghai, China 5 Karlsruhe Institute of Technology (KIT), Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany 6 Department of Drug and Health Sciences, University of Catania, V.le A. Doria, 6, 95125 Catania, Italy 7 Department of Industrial Engineering, Alma Mater Studiorum—University of Bologna, Via Zamboni, 33-40126 Bologna, Italy *Corresponding author. Department of Drug and Health Sciences, University of Catania, V.le A. Doria, 6, Catania 95125, Italy. E-mail: francesco.pappalardo@unict.it ‡Authors are listed in alphabetical order; being this a consensus paper, all authors should be considered first authors with equal roles. Abstract The rapid integration of machine learning (ML) predictors into in silico medicine has revolutionized the estimation of quantities of interest that are otherwise challenging to measure directly. However, the credibility of these predictors is critical, especially when they inform high-stakes healthcare decisions. This position paper presents a consensus statement developed by experts within the In Silico World Community of Practice. We outline 12 key statements forming the theoretical foundation for evaluating the credibility of ML predictors, emphasizing the necessity of causal knowledge, rigorous error quantification, and robustness to biases. By comparing ML predictors with biophysical models, we highlight unique challenges associated with implicit causal knowledge and propose strategies to ensure reliability and applicability. Our recommendations aim to guide researchers, developers, and regulators in the rigorous assessment and deployment of ML predictors in clinical and biomedical contexts. Keywords: machine learning credibility; in silico medicine; causal knowledge assessment; regulatory science; error quantification; bias robustness Introduction The advent of machine learning (ML) has ushered in a new era in biomedical engineering and in silico medicine, offering unprecedented capabilities in modeling complex biological systems and predicting clinically relevant outcomes. ML predictors have emerged as powerful tools for estimating quantities of interest (QIs) that are difficult or impossible to measure directly, such as disease risk, treatment efficacy, or physiological parameters within the human body [1–7] . These advancements promise to transform healthcare by enabling personalized medicine, optimizing therapeutic strategies, and improving patient outcomes. However, with great potential comes significant responsibility. The credibility of ML predictors is paramount, as inaccurate or unreliable predictions can lead to misdiagnosis, inappropriate treatments, and harm to patients [8 , 9] . The reliance on data-driven models introduces unique challenges. ML predictors often function as “black boxes”, lacking transparency in how inputs are transformed into outputs. They depend heavily on the quality and representativeness of the training data and may inadvertently capture biases or spurious correlations. Furthermore, the absence of explicit causal relationships complicates the validation and reg

FAQs

Why is credibility assessment necessary for machine learning predictors in in silico medicine?

ML predictors can inform high-stakes healthcare decisions, where inaccurate or unreliable predictions may harm patients. Credibility assessment helps ensure results are trustworthy for clinical use.

What core elements does the consensus statement emphasize for evaluating credibility?

The statement highlights the need for causal knowledge, rigorous error quantification, and robustness to biases when assessing ML predictors.

How does the document position ML predictors compared with biophysical models?

It discusses unique challenges in ML arising from implicit causal knowledge and focuses on strategies to support reliability and applicability compared with biophysical modeling approaches.

Aug 4, 2026

Upload by Levi