Enhancing Healthcare Facility Management through Non-intrusive Occupancy Detection: A Comparative Study of Machine Learning Algorithms

Optimizing patient environments in healthcare is crucial for improving care quality and operational efficiency. This study explores the application of machine learning algorithms for non-intrusive room occupancy detection in healthcare settings, which is essential for creating responsive, patient-centered environments. Using sensors to monitor temperature, light, sound, and CO2 levels, we applied Random Forest (RF), k-nearest Neighbors (k-NN), Support Vector Machines (SVM), and Decision Trees (DT) to predict room occupancy. The study focused on evaluating the accuracy and privacy-preserving capabilities of these algorithms. The results demonstrated that RF, k-NN, and DT performed exceptionally well, achieving high precision, recall, and F1 metrics scores, highlighting their potential for error-free occupancy prediction in healthcare environments. In contrast, while typically robust, the SVM algorithm exhibited slight variations in precision for certain classes, underscoring the importance of algorithm-specific considerations in different healthcare settings. Feature importance analysis revealed that light-based sensors and CO2 levels were critical in determining room occupancy, aligning with existing literature emphasizing environmental factors' significance in patient care settings. This study underscores the valuable role of machine learning in enhancing healthcare environments and advocates for the integration of such technologies into healthcare management systems. This integration promises significant improvements in patient care and resource optimization. Future research should focus on examining the impact of these technologies on patient outcomes and healthcare efficiency.