Odette Rios-Ibacache | Search arXiv

arXiv2026

MedHal: a Synthetic Dataset for Medical Hallucination Detection

Hallucination, the generation of non factual content by AI systems, poses serious risks in medical contexts, where errors can directly affect patient outcomes. We present MedHal, a large-scale dataset specifically designed to assess capabilities and train models on the task of hallucination detection in medical texts. Current hallucination detection methods face significant limitations when applied to specialized domains like medicine, where they can have disastrous consequences. MedHal addresses this issue by incorporating diverse medical text sources and tasks covering both intrinsic and extrinsic hallucinations, and by providing a substantial volume of data samples suitable for training medical hallucination detection models. We demonstrate MedHal's utility by training and evaluating a baseline medical hallucination detection model, showing improvements over general-purpose hallucination detection approaches. This resource enables more efficient evaluation and training of medical text generation systems while reducing reliance on costly expert review, potentially accelerating the development of medical AI research.

cs.CL↗

arXiv2025

Quantification of head and neck cancer patients' anatomical changes during radiotherapy: prediction of replanning need

Head and neck cancer (HNC) patients who undergo radiotherapy (RT) may experience anatomical changes during treatment, compromising the validity of the initial treatment plan, necessitating replanning. However, replanning disrupts clinical workflows, creating a stressful environment. Currently, no standardized method exists to determine the total amount of anatomical change that necessitates replanning. This project aimed to create metrics to describe anatomical changes HNC patients may experience during RT and develop machine learning (ML) models to predict RT replanning. We included a cohort of 150 HNC patients treated at the McGill University Health Centre. Based on the shape of the RT structures, we created metrics and developed an extraction pipeline, called HNGeoNatomyX, to automatically calculate them. A univariate metric analysis using linear regression was conducted to obtain the rate of change of each metric. We also obtained the relative variation of each metric between the pre-treatment scan and the fraction at which replanning was requested. Fraction-specific ML models (models that incorporated information available up to and including the specific fraction) for fractions 5, 10, and 15 were built using the metrics, clinical data, and feature selection techniques. To estimate models' performance, we used a repeated stratified 5-fold cross-validation resampling technique and the Area Under the Curve (AUC) of the Receiver Operating Characteristic (ROC) curve. The best specific multivariate models for fractions 5, 10, and 15 yielded testing scores of 0.82, 0.70, and 0.79, respectively. Our models early predicted replanning for 76% of the true positives. The created metrics have the potential to characterize and distinguish which patients will necessitate RT replanning. They show promise in guiding clinicians to evaluate RT replanning for HNC patients and streamline workflows.

physics.med-ph↗