Revolutionizing hospital supply chain: machine learning for predictive risk assessment

Purpose The purpose of this study is to identify and prioritize the risks associated with healthcare supply chain with a focus on hospital’s internal supply chain. Design/methodology/approach The study presents a machine learning model for hospital supply chain risk assessment. Through previous literature and interviews, 35 risk factors were identified, which were then used to devise a questionnaire aimed at assessing the probability and impact of these risk factors. A total of 140 responses were gathered from subject matter experts. Six machine learning algorithms, namely, Naïve Bayes (NB), Random Forest (RF), K-Nearest Neighbors (KNN), Logistic Regression (LR), Support Vector Machines (SVM) and Artificial Neural Networks (ANN) were used in this study. Techniques including holdout validation and k-fold cross-validation were used for evaluating the six classifiers. Findings Results from the feature subset selection process revealed that the most significant variables included “black market,” “delay in replacement of defective products,” “delay in delivery of imported products,” “defaulted suppliers,” “mismanagement of expired medicines,” “lack of planning,” “ban on import of raw materials” and “lack of budget.” While holdout validation results showed that Naïve Bayes achieved highest classification accuracy of 85.7%, the k-fold cross-validation indicated that RF, ANN and KNN outperformed NB, LR and SVM in terms of accuracy, highlighting their superior generalization capability. Originality/value The study fills an important research gap by providing valuable and novel insights for healthcare institutes and policymakers in making informed decisions regarding risk mitigation strategies, resource allocation and operational planning. By shifting their focus on the critical risk factors, stakeholders can prioritize their efforts and resources more effectively.