[Application of artificial intelligence for an early comparison of efficacy between new cancer drugs.]

INTRODUCTION:The clinical choice among recently approved cancer drugs is burdened by the absence of direct comparisons in terms of efficacy across these new agents. In this article we present the IPDfromKM method, an artificial intelligence (AI) application that aims to facilitate the analyses on efficacy based on secondary data. METHODS:Seven therapeutic areas were selected in which at least three new agents were recently approved. Kaplan-Meier curves of related clinical trials were digitized. Then, the IPDfromKM method was employed to reconstruct patient-level survival data. This information allowed us to compare selected agents in each therapeutic area and to rank them in terms of efficacy. RESULTS:We identified the most effective treatment in each of the seven selected therapeutic areas. In two cases, immunotherapies, sharing similar mechanisms of actions, were compared highlighting the most effective one. In the remaining cases, our comparison included also the standard of care, which proved to be superior to new agents in patients with osteosarcoma. DISCUSSION:When randomized clinical trials are not available, indirect comparisons can be a valuable source of information. The experience described herein recommends the use of a new method endowed by two important advantages: remarkable speed of analysis and free access to computational tools. In assessing the place in therapy for newly developed agents, this approach can further promote the application of evidence-based principles.