The integration of artificial intelligence (AI) into clinical trials promises to revolutionize biomedical research by enhancing efficiency, reducing costs, and accelerating the delivery of novel therapies. However, the discourse surrounding this integration is often characterized by a significant gap between proclaimed potential and the complex realities of real-world implementation. This critical review moves beyond a descriptive survey of AI applications to provide a rigorous, evidence-based synthesis of the field. It scrutinizes the methodological underpinnings of AI across the clinical trial lifecycle, from protocol design and patient recruitment to data analysis and outcome prediction. The analysis delves into the comparative performance of specific AI techniques, such as deep learning and natural language processing, grounding the discussion in quantitative metrics and highlighting the critical trade-offs between accuracy, interpretability, and scalability. Claims of efficiency gains and cost reduction are critically evaluated against a backdrop of peer-reviewed economic analyses and documented implementation failures. This review systematically deconstructs the profound challenges that define the current landscape: the pervasive issue of algorithmic bias and its implications for health equity, the “black box” dilemma of model transparency, and the persistent implementation gap between controlled studies and diverse clinical settings. Furthermore, it dissects the unresolved technological hurdles of emerging decentralized systems like federated learning and blockchain, and confronts the pressing ethical and governance debates surrounding data ownership, consent, and legal liability. By weaving together disparate and often conflicting findings into a cohesive critical narrative, this paper identifies crucial research gaps and proposes a forward-looking agenda. It serves as an essential analytical resource for researchers, clinicians, regulators, and ethicists, aiming to foster a more nuanced understanding of the turbulent but promising path toward the responsible and impactful deployment of AI in clinical trials.
Lim et al. (Fri,) studied this question.