Machine Learning to Identify Physician Actions Associated with Patient Experience of Compassion
A total of 1025 patients completed the questionnaire (717 from site 1; 308 from site 2). The action with the strongest association with greater compassion was "Listen carefully to what you had to say" [β = 5.67 (95% CI 3.36 to 7.98)]. The LASSO algorithm identified nine actions as the best predictors, and this model explained roughly a third of variation in compassion scores: r<sup>2</sup> = 0.34 in the derivation cohort and r<sup>2</sup> = 0.32 in the validation cohort. Three behaviors had a stronger association with compassion scores among women (compared to men) and five behaviors among Black (compared to White) individuals. CONCLUSIONS AND RELEVANCE: These results show specific physician behaviors are associated with patients' experience of compassion. Further research is needed to develop interventions to foster these behaviors and test prospectively whether they improve patients' experience of compassion.
Authors
- Clifford M. Marks
- Amanda Scudder
- Lauren B. Remboski
- Nathan Ivan Shapiro (ORCID: https://orcid.org/0000-0003-3508-4390)
- Andrew Nyce
- Christopher W. Jones (ORCID: https://orcid.org/0000-0001-9704-9094)
- Stephen Trzeciak (ORCID: https://orcid.org/0000-0002-7048-3330)
- Clifford Gaines
- Adrian D. Haimovich (ORCID: https://orcid.org/0000-0002-4106-7055)
- Brian W. Roberts (ORCID: https://orcid.org/0000-0002-7690-997X)
- P. Baptista
Institutions
- Beth Israel Deaconess Medical Center (US)
- Cooper University Hospital (US)
- Cooper Medical School of Rowan University (US)
- Cooper University Health Care (US)
Publication Details
- Journal
- Journal of General Internal Medicine
- Published
- 2025-10-21
- DOI
- https://doi.org/10.1007/s11606-025-09914-8
- Primary Topic
- Empathy and Medical Education
- Type
- article
- Field-Weighted Citation Impact
- 0.00