![[2602.22285] Early Risk Stratification of Dosing Errors in Clinical Trials Using Machine Learning](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.22285] Early Risk Stratification of Dosing Errors in Clinical Trials Using Machine Learning
Summary
This study presents a machine learning framework for early risk stratification of dosing errors in clinical trials, utilizing pre-initiation data to predict error likelihood.
Why It Matters
The ability to predict dosing errors before clinical trials begin can significantly enhance patient safety and trial efficiency. By employing machine learning, researchers can proactively manage risks, ensuring better outcomes and resource allocation in clinical research.
Key Takeaways
- Developed a machine learning framework for risk stratification of dosing errors in clinical trials.
- Utilized a dataset of 42,112 clinical trials to train models, achieving an AUC-ROC of 0.862.
- Probability calibration was crucial for creating interpretable risk categories.
- The late-fusion model combined structured and unstructured data for improved performance.
- The framework supports proactive quality management in clinical research.
Computer Science > Machine Learning arXiv:2602.22285 (cs) [Submitted on 25 Feb 2026] Title:Early Risk Stratification of Dosing Errors in Clinical Trials Using Machine Learning Authors:Félicien Hêche, Sohrab Ferdowsi, Anthony Yazdani, Sara Sansaloni-Pastor, Douglas Teodoro View a PDF of the paper titled Early Risk Stratification of Dosing Errors in Clinical Trials Using Machine Learning, by F\'elicien H\^eche and 4 other authors View PDF HTML (experimental) Abstract:Objective: The objective of this study is to develop a machine learning (ML)-based framework for early risk stratification of clinical trials (CTs) according to their likelihood of exhibiting a high rate of dosing errors, using information available prior to trial initiation. Materials and Methods: We constructed a dataset from this http URL comprising 42,112 CTs. Structured, semi-structured trial data, and unstructured protocol-related free-text data were extracted. CTs were assigned binary labels indicating elevated dosing error rate, derived from adverse event reports, MedDRA terminology, and Wilson confidence intervals. We evaluated an XGBoost model trained on structured features, a ClinicalModernBERT model using textual data, and a simple late-fusion model combining both modalities. Post-hoc probability calibration was applied to enable interpretable, trial-level risk stratification. Results: The late-fusion model achieved the highest AUC-ROC (0.862). Beyond discrimination, calibrated outputs enabled robust...
