Fig. 1.

Workflow for ML prediction model development. Pharmacogenomic data from cell lines, patient-derived xenografts (PDXs), and patient materials are ideal for ML model development. Data from different sources are preprocessed and then divided into training (including cross-validation) and test groups. The training dataset is used to build and validate the prediction model, while the test dataset is used for testing the model’s accuracy and precision. To develop a prediction model for clinical use, vigorous preclinical assessment is required that can be performed using cell lines, PDXs, and patient materials that have not been used for model development. Additionally, the efficacy of predicted drugs must be tested for disease-specific preclinical models. Finally, both the model and predicted drug will undergo a clinical trial.