Association between AI-driven conversational agents and physician-patient interaction quality during outpatient...

This study aimed to use a propensity score matching (PSM) design to examine the association between artificial intelligence (AI)-driven conversational agents (CAs) and physician-patient interaction quality during outpatient consultations.

We used the Chinese version of the Consultation and Relational Empathy Measure to survey the patients' perceived quality of physician-patient interactions during outpatient consultations, involving 419 adult residents who received outpatient services from China's tertiary public hospitals. Propensity score matching was first conducted to organize the sampled population into the treated and control groups based on the demographic and visit covariates, and the average treatment effect on the treated (ATT) was further calculated to estimate the causal association between the AI-driven CAs and physician-patient interaction quality.

Overall, the PSM results showed a positive causal association of the AI-driven CAs with the physician-patient interaction quality. Specifically, the ATT estimate results showed that the treated residents gave significantly higher scores than the control residents in the total perceived physician-patient interaction quality score (ATT=2.987, Z=2.92, P=0.003) and its 8 items at the 5% confidence level. The sensitivity analysis results further showed that when the $\gamma$ increased to greater than 2, the ATT estimate results remained significant (P<0.001), indicating the ATT estimate results were not sensitive to the hidden bias.

Our findings will help to confirm the association between the AI-driven CAs and physician-patient interaction quality, and also offer valuable guidance for policy makers and hospital managers in promoting the adoption of the AI-driven CAs to continuously improve the physician-patient interaction quality during outpatient consultations.