PredictCan-MIND v2.1.2 was developed and trained using a comprehensive reference dataset comprising more than 40 drugs and data from over 2,000 individuals, including both males and females aged 18 to 70 years. Importantly, the reference set includes drugs spanning different pharmacological and therapeutic classes and covering the full spectrum of DILI.  This diversity is a key strength of the model, as it exposes the algorithm to a wide spectrum of drug-related characteristics and patient profiles rather than restricting its development to a narrow drug class or a highly homogeneous population.

This breadth of pharmacological and demographic representation enhances the platform’s predictive robustness, enabling it to capture interindividual variability that underlies idiosyncratic DILI susceptibility.