Reducing Patient Waiting Time at Public Hospitals Through Predictive Modeling

Reducing patient waiting times in public hospitals is a critical challenge that significantly impacts patient satisfaction, healthcare outcomes, and operational efficiency. Efficient, effective, and timely patient care contributes toward patient service satisfaction. This is not always true for most healthcare institutions. This paper investigates the use of predictive modeling in addressing patient waiting time by leveraging historical data from electronic records at five selected public health centers in Harare, Bulawayo, and Gweru between October 1 and 31 December 2023. The objective was to identify a predictive model that could effectively classify the queue data into two categories: delayed (Yes) or not delayed (No). Logistic Regression, Decision Tree, Random Forest, and XGBoost models were trained using Stratified K-Fold cross-validation to determine the model with the best fit on the original data. The models were trained using variables such as queue position, wait time, service time, location of the health center, and weekday as predictors. XGBoost emerged as the top model, achieving the highest ROC AUC of 0.63, compared to the other models which ranged from 0.56 to 0.6. The selected model was then evaluated using F1 Score, ROC Curve, and Precision-Recall Curve. Results showed a high F1 Score of 0.702, indicating the model's ability to distinguish between the positive (Delayed) and negative (Not Delayed) classes. The area under the ROC curve (AUC) of 0.63 further quantified the model's overall ability to discriminate between these classes. Additionally, the area under the Precision-Recall curve (AUC-PR) of 0.64 suggests that the model performs well in balancing precision and recall, effectively identifying most actual delayed services (high recall). Accurately predicting patient waiting times not only enhances patient care experience but also supports healthcare providers in delivering timely and efficient care. Designers of hospital programs can analyze patterns, and identify peak times, predict high-delay patient arrivals, and suggest real-time staffing adjustments. Future work will focus on refining the models and integrating them into hospital management systems for broader implementation.