The success rate of pharmaceutical Research & Development (R&D) is much lower compared to other industries such as micro-electronics or aeronautics with the probability of a successful clinical development to approval in central nervous system (CNS) disorders hovering in the single digits (7%). Areas covered: Inspired by adjacent engineering-based industries, we argue that quantitative modeling in CNS R&D might improve success rates. We will focus on quantitative techniques in early clinical development, such as PharmacoKinetic-PharmacoDynamic modeling, clinical trial simulation, model-based meta-analysis and the mechanism-based physiology-based pharmacokinetic modeling, and quantitative systems pharmacology. Expert commentary: Mechanism-based computer modeling rely less on existing clinical datasets, therefore can better generalize than Big Data analytics, including prospectively and quantitatively predicting the clinical outcome of new drugs. More specifically, exhaustive post-hoc analysis of failed trials using individual virtual human trial simulation could illuminate underlying causes such as lack of sufficient functional target engagement, negative pharmacodynamic interactions with comedications and genotypes, and mismatched patient population. These insights are beyond the capacity of artificial intelligence (AI) methods as they are many more possible combinations than subjects. Unlike 'black box' approaches in AI, mechanism-based platforms are transparent and based on biologically sound assumptions that can be interrogated.
Keywords: PB–PK modeling; Quantitative systems pharmacology; clinical trial; failure analysis; virtual patient.