A Classification Algorithm for Managing Electronic Medical Reports to Ensure Equitable Physician Workload Distribution in Healthcare Facilities

This research investigates the escalating issue of effectively managing electronic medical records (EMRs) within high-paced healthcare settings, where the disproportionate allocation of documentation responsibilities can contribute to physician exhaustion, hinder timely patient care, and elevate the likelihood of errors. We conceptualize the distribution of EMRs as a load-balancing challenge: the aggregate workload of each physician encompasses both direct clinical services provided and the time needed to handle assigned EMRs. To quantify EMR processing duration, we represent each record in terms of its length (measured by the number of paragraphs) and a complexity score derived from its clinical content. In pursuit of a fair distribution of EMRs, we propose a series of classification-based load-balancing algorithms. Initially, records are classified into five categories based on their processing duration and complexity, subsequently assigned to physicians in accordance with their current workload and historical processing capabilities. Among the methodologies presented are adaptations of decision-tree segmentation, genetic programming with threshold settings, and an innovative Classification of Three Groups of Physicians (CTGP) technique that flexibly modifies assignment criteria as workloads fluctuate. We assessed these algorithms using a benchmark of 1,250 synthetic EMR cases, evenly stratified across the five categories. The evaluation criteria included assignment accuracy (i.e., the degree to which each physician’s workload approximated the optimal balanced condition), average optimality gap, and execution duration. The CTGP algorithm yielded the most favorable outcomes, attaining an assignment accuracy of 88.0%, an average optimality gap of 0.05 concerning the ideal balance, and an average execution time of 0.004 seconds per instance. These results illustrate that classification-driven load balancing can proficiently allocate EMR responsibilities, presenting a scalable remedy for contemporary electronic health record frameworks. Future investigations will consider the validation of real-world clinical data and its integration with dynamic scheduling systems.