# Determinants of Chinese physicians’ engagement in narrative medicine: a comprehensive SEM-ANN analysis

## TL;DR

This study explores what influences Chinese doctors to engage in narrative medicine, using a combination of theories and data analysis to identify key factors and patterns.

## Contribution

A hybrid SEM-ANN model is introduced to analyze behavioral intentions toward narrative medicine, revealing nonlinear interactions and distinct clinician profiles.

## Key findings

- Perceived behavioral control was the strongest predictor of narrative medicine practice intentions.
- Four distinct clinician profiles were identified based on psychological motivation and behavioral intention coupling.
- Profile distribution varied significantly across hospital levels and departments.

## Abstract

The implementation of narrative medicine (NM) holds significant implications for multiple stakeholders in healthcare, necessitating investigation into clinicians’ behavioral intention (BI) toward NM practice. This study employs the Theory of Planned Behavior (TPB) to identify predictors of Chinese clinicians’ NM practice intentions, extending the theoretical framework through incorporation of external variables—Perceived Organizational Support (POS) and Perceived Informal Organizational Support (PIOS).

Data collected from 855 Chinese clinicians validated the theoretical model. The hybrid model of Structural Equation Modeling (SEM) and Artificial Neural Network (ANN) demonstrated superior predictive accuracy, successfully capturing both linear associations and nonlinear interactions among variables. Phase I utilized SEM to identify intention predictors, while Phase II applied ANN to rank predictors’ relative importance. Furthermore, conducting K-means clustering (optimal K = 4 determined by Elbow Method and Silhouette Coefficient) using ATT, SN, PBC, and PIOS as inputs to establish a “Psychological Motivation-Behavioral Intention” coupling profile. Chi-square tests examined distribution differences across organizational contexts.

Perceived behavioral control (PBC) was the strongest BI predictor, followed by PIOS, subjective norms (SN), and attitudes (ATT), with POS being non-significant. Four distinct profiles emerged: “Moderately Engaged, Moderately Intent” (Cluster 1), “Attitudinally Compliant, Structurally suppressed” (Cluster 2), “Cognitively Engaged, Moderately Intent” (Cluster 3), and “Fully Engaged, High-Intent” (Cluster 4). Distribution varied significantly across hospital levels (χ2 = 22.297, p = 0.001) and departments (χ2 = 26.240, p = 0.036), with Cluster 2 predominant in emergency/pediatrics and Cluster 4 dominant in surgery.

ATT, SN, PBC, and PIOS positively influenced physicians’ NM behavioral intention. The motivation-intention coupling profiles reveal heterogeneity in practice potential, offering theoretical-practical insights for targeted interventions.