All Times EDT
Keywords: Missing data, Precision medicine, Ranking procedures, Artificial intelligence
A great deal has been accomplished in the innovative use of statistical design and analysis techniques of the safety and effectiveness/efficacy of pharmaceutical drugs, biologics and medical devices. Statistical pioneers in industry, regulatory statistical scientists and academicians, often working in harmony, have driven these innovations. The retrospective focus of this presentation is the progress in three areas: (1) missing data (tipping point and estimands), (2) design and analysis in precision medicine and the evaluation of diagnostic medical tests (especially for companion diagnostics) and (3) ranking procedures with applications to analysis of multiple endpoints for a composite and to benefit-risk assessment. The deluge of big data will result in increased use of artificial intelligence/machine learning techniques for many imaging and other diagnostic procedures, a data science challenge for statisticians. The future holds a greater use of: (1) Bayesian and adaptive approaches, facilitated by simulations; (2) causal models, including propensity scores, with real world evidence implications, leveraging electronic health records and other data sources for the pre-market and post-market; (3) patient input and preferences. The COVID-19 pandemic has highlighted the heightened importance of Bayesian designs, platform trials from oncology and the home use of therapeutic and diagnostic products.