OBJECTIVES: High placebo response rates can reduce the difference between active treatment and control groups in clinical trials and are a challenge in drug development. There is currently no consensus on how to statistically control for the placebo effect. Researchers can now use artificial intelligence (AI) methods, such as machine learning and natural language processing, to identify and model predictors of the placebo response at an individual patient level.

This study provides an overview of AI methods used, research goals, and indications; and explores the implications for clinical research and practice.

METHODS: This review was based on a search of MEDLINE, Embase, and PsycInfo databases via the ProQuest platform using title/abstract keywords and subject heading synonyms for AI, machine learning, natural language processing, and placebo response/effect. Studies evaluating AI methods to investigate the placebo response were eligible for inclusion.

RESULTS: The initial search identified 92 records; two additional articles were identified via a bibliography review of the selected articles. After deduplication, 84 records remained; of these, 19 records met the eligibility criteria for inclusion in this review.

The majority of the included studies related to major depressive disorder and other psychiatric conditions and used various machine learning methods to study the placebo response through predictors/modulators of placebo versus medication response. One study, in chronic pain, used natural language processing. The suggested uses for these predictors were to statistically account for unbalanced allocation in clinical trials and to guide treatment.

CONCLUSIONS: AI methods can predict patients who are likely to respond to placebo. Statistically controlling for the placebo effect could provide a more accurate estimate of treatment effect sizes in clinical trials. In medical practice, patients likely to respond to placebos could benefit from less-intensive treatments. Data privacy is essential, as those likely to respond to placebo might experience stigmatization.