Assessment of process management and maturity in public hospitals: an explainable machine learning approach

Abstract

BACKGROUND: In healthcare systems, hospitals are considered as the key organizations. Their activities are shaped by three main axes: medical, academic, and managerial. Because of their importance, management concerns have become more important in recent years. In order to face the growing performance and innovation challenges, public hospitals are recently adopting new managerial practices, based on more dynamic decision making. In this context, this research we aim to evaluate the current process management maturity in Moroccan Public Hospitals (MPH), and to identify the main organizational dimensions associated with higher observed levels of process maturity in this context. METHODS: To assess the level of maturity by process in MPH, we adopted the Business Process Orientation Maturity Model, based on seven axes. Data were collected using a questionnaire, and initial inspection showed that <10% of responses were missing. These incomplete entries were removed, resulting a final dataset of 154 complete observations. The analysis was conducted on three phases. The first step was ranking hospitals based on the average maturity scores for each axis. The second step involved applying three machine learning (ML) algorithms: Logistic Regression, Random Forest, and XGBoost (Extreme Gradient Boosting), to classify hospitals according to their observed maturity level and identify priority areas associated with higher maturity. Finally, we applied SHAP (SHapley Additive exPlanations) method to interpret the decisions made by the ML algorithms and identify the most important axes associated with higher maturity level. RESULTS: The main results of this study show that the current maturity level of prefectural and provincial hospitals is below 2. Meanwhile, university hospitals (UH) recorded a current maturity level above 2. The application of explainable artificial intelligence methods, particularly SHAP, revealed that the three algorithms converged on the same priority areas. Risk Management and Compliance Control, and Involvement of Top Management and Strategic Alignment emerged as the most axes associated with maturity level ≥2. In addition, Optimization: Action Plan and Performance Measurement and Socio-economic and Operational Performance were also marked as important. CONCLUSION: The study highlighted the challenges associated with modernizing hospital management in Morocco in a context marked by organizational complexity and growing demands in terms of performance and quality of care. In line with the process management approach, we used maturity model to assess the positioning of MPH on seven key areas. The empirical results reveal a significant gap between UH and first- and second-level hospitals. Beyond measurement and prediction, this study is positioned as a proof-of-concept illustrating how explainable ML can complement process maturity assessment frameworks, especially in Moroccan Public Hospitals (MPH).