Retrospective observational study applying reinforcement learning and behavioral science to mammography: insights for equitable preventive care
In brief
One in five women opening tailored emails completed mammography screening
In this retrospective study, 20.3% of women who opened at least one tailored email attended a mammogram; 23.7% scheduled one. Capability-focused messages were generally linked to higher uptake, while opportunity-focused messages were linked to lower uptake, with some differences across demographic groups. Because the study was observational and assessed only engaged patients, it cannot show that messaging caused screening or narrowed disparities.
- Journal
- Annals of behavioral medicine : a publication of the Society of Behavioral Medicine (Q1)
- Published
- 7 January 2026
- Study design
- Cohort / observational study
- Evidence level
- Level 3, Low (CEBM 3b)
- Authors
- Emily Frith, Sarah Deedat
- PMID
- 42847375
- DOI
- 10.1093/abm/kaag024
Why clinicians should know about it
- Picked for Health Informatics (top studies of the week, 11 October 2026): Reinforcement learning nudges improve mammogram uptake
Abstract
BACKGROUND: In the United States, underserved women face persistent healthcare disparities, making mammography screening and equity public health priorities. PURPOSE: This retrospective observational study assessed how tailored messaging influenced mammogram scheduling and attendance, determined which message types were most effective, and explored differences across key demographic subgroups. METHODS: We employed Precision Nudging using reinforcement learning and behavioral science to tailor email messages informed by the Capability, Opportunity-Behavior (COM-B) model. Messages were sent to 156 954 women overdue for screening. Mammogram scheduling and attendance were documented via electronic health records. Analyses focused on the engaged patient population who opened at least one email (N = 106 616; 67.9%). Logistic regressions examined whether COM-B content in the first and last message predicted scheduling and attendance, testing interactions for age, race, income, insurance, and education. RESULTS: Among engaged patients, 23.7% scheduled and 20.3% attended a mammogram. Capability-focused messaging was generally associated with higher overall uptake, while opportunity messaging consistently predicted lower scheduling and attendance rates. Equity analyses revealed nuanced engagement patterns among underserved groups. Underrepresented racial/ethnic minority women had higher odds of completing screening than White women. Older age predicted lower attendance, with early messages addressing motivation predicting lower attendance than capability messages. Final intervention messages addressing motivation were more effective for low- and high-income women than middle-income women and more effective for women without private insurance. CONCLUSIONS: These findings suggest that tailoring messages to patients' needs and context encourages mammogram uptake and may help reduce disparities in women's health.
Abstract as published, via PubMed.
For healthcare professionals. The summary is generated by AI from the published abstract, and the evidence level is assigned automatically from the study design on the Oxford CEBM hierarchy. Neither is medical advice. Read the full paper before changing practice.