With machine learning rapidly becoming a key part of various scientific fields, understanding the role of machine learning for questions in program evaluation, comparative effectiveness, and precision medicine is important.

As co-editors of the journal Biostatistics, associate professor of health care policy (biostatistics) Sherri Rose, PhD, and professor of biostatistics at the Erasmus Medical Center Rotterdam Dimitris Rizopoulos, PhD, organized a series of commentaries on machine learning for causal inference.

The five commentaries were gathered from scholars with expertise in statistics, computer science, epidemiology, health economics, policy, and law.

The first discusses the prevalence of structural racism in machine learning and the importance of assessing and eliminating algorithmic bias. The next commentary addresses the dangers of a lack of generalizability and the importance of generalizability standards.

The next two commentaries deal with estimation. The first investigates procedures that rely on semiparametric theory in order to build machine learning-based effect estimators that can produce causal estimates. The next explores the estimation of individual-level treatment rules when machine learning is employed in precision medicine and identification of individualized treatments.

The last commentary speaks on regulatory and policy issues in the integration of machine learning in health care. Safety and effectiveness are of key importance in decision support and testing as well as medical devices with machine based-learning software.

As described in their editorial, Rose and Rizopoulous hope that these commentaries create conversations across disciplines and promote thoughtful consideration when pursuing machine learning.