Machine Learning Approaches for Phase 2 Clinical Trial Outcome Prediction in Oncology: A Comprehensive Review and the BioClinical Fusion AI Framework

It has been taking more than a decade to develop a drug and get it into the regulation. Among this, Phase II clinical trials are the critical stage, which has been provides 60% - 70% of failure rate due to the lack of efficiency and toxicity of drug. Moreover, the repeated failure rate leads to the consumption of cost and time of pharmaceutical industries. This review examines the integration of Artificial Intelligence with clinical trials to predict the success or failure rate of phase II clinical trials. This review is mainly focused on the area of Bioinformatic which provides with the use of computational tools and machine learning approach and likely the area of pharmaceutical science provides with the five set of data such as Biological, Mechanism of action of drug, clinical, preclinical and historical databases. The development of AI predictive frame-work is the fusion of these five datasets which is termed as multi-layered bio-clinical fusion AI. Various machine learning models such as XGBoost, logistic regression, and random forest have been widely applied in this domain which comprehensively provides the predictive score of the Phase II clinical trials by generating the F2 score and also validate the clinical trials result by comparing traditional drug with competitive drug. The reviewed studies suggest that AI-based approaches have the potential for the pharmaceutical industries in the development of drug by reducing the expenses and time consumption. Keywords: Phase II clinical trials, Artificial Intelligence, Machine learning, bio-clinical fusion, F2 score.