Prediction of drug approvals in oncology using explainable artificial intelligence

Oncology is characterized by vigorous drug development, including the introduction of novel therapeutic approaches. However, the success rate of clinical trials in oncology is markedly lower than that in other disease areas. Previous studies have developed machine learning models to predict drug approval by utilizing drug development and clinical trial data to optimize drug development portfolios. However, these models do not adequately quantify trial design information specific to oncology, which is considered insufficient for portfolio optimization in this domain. In this study, we developed a machine learning model that incorporates oncology-specific trial design information to predict drug approval. The prediction models we constructed achieved an area under the receiver operating characteristic curve of 0.86 to 0.98 and demonstrated high accuracy in prospective validation. Furthermore, our findings suggested that features related to trial duration, the number of enrolled patients and the number of products launched by the sponsor contributed to the prediction of drug approval. By utilizing the predictive model established in this study, estimates of the approval probabilities for drugs in the oncology pipeline can be provided, thereby aiding the strategic planning of drug development.