Influencing factors and mediating effects of advanced medical care plan readiness in young and middle-aged patients with advanced cancer based on machine learning.

OBJECTIVE: To examine the factors influencing advance care planning (ACP) readiness and its correlation with disease benefit perception and social support perception among young and middle-aged patients with advanced cancer, as well as to analyze the mediating effect of disease benefit perception on the relationship between social support and ACP readiness.

METHODS: A total of 432 young and middle-aged in patients with advanced cancer from a tertiary cancer hospital in Guangxi were selected through convenience sampling from October 2024 to September 2025. The patients were assessed using a general information questionnaire, an advanced care plan readiness scale, a disease benefit sense scale, and a comprehension social support scale. First, the variables were assessed using a combination of LASSO regression and random forest techniques. Subsequently, the selected variables were incorporated into a multiple linear regression model to examine the determinants of ACP readiness, with the robustness of the findings evaluated through the Bootstrap resampling method. On this basis, mediation analysis was used to test the mediating effect of disease benefit perception on the understanding of social support and ACP readiness, and gender subgroup analysis was conducted.

RESULTS: The readiness score of advance medical care plan for young and middle-aged advanced cancer patients was (76.38 ± 14.40), the disease benefit sense score was (72.50 ± 14.65), and the comprehension of social support was (52.23 ± 12.99). The outcomes of machine learning feature selection indicated that social support, perception of disease benefits, and awareness of advance care planning (ACP) significantly influence ACP readiness. Multiple linear regression analysis showed that ACP readiness was significantly reduced (β = -6.43, P < 0.001), disease benefit (β = 0.33, P < 0.001) and social support (β = 0.49, P < 0.001) were independent positive influencing factors of ACP readiness. The findings from the mediation analysis indicated that the perception of benefit derived from the disease served as a partial mediator between perceived social support and readiness for advance care planning, exhibiting an indirect effect of 0.126 (95% CI: 0.079-0.182), which constitutes 20.0% of the overall effect. The subgroup analysis showed that this mediating effect was significant in both male and female patients and was slightly higher in males than in females.

CONCLUSION: The readiness for ACP in young and middle-aged patients with advanced cancer is moderate. Recognizing that social support not only directly influences the preparedness of ACP programs but also serves as a partial mediator through the perceived benefits derived from the disease. Healthcare professionals ought to enhance the development of patients' social support networks within clinical settings and focus on elevating the perceived benefits associated with their conditions. This approach aims to foster patients' inclination to engage in ACP and to refine the quality of care provided at the end of life.