Abstract:
Currently, the management of blood glucose for inpatients outside the endocrinology department lacks standardization. Common issues include inconsistent diagnosis and treatment of diabetes among hospitals at different levels, inefficient referral procedures, and a shortage of data sharing and intelligentization. To tackle these issues, this study constructed a data production model via cross-domain standardized data collection, structured input, and intelligent annotation. A dynamic three-color (red-yellow-green) hierarchical medical labeling system for diabetes was established, with the generation of a high-quality diabetes “three-color label” dataset, and the integrated regional blood glucose management information system. Within hospitals, the information system automatically identifies hyperglycemic patients in non-endocrinology inpatient departments, initiating consultation reminders and blood glucose management, and set up a “virtual ward” for these patients. The endocrinology specialty teams offered cross-departmental blood glucose management services. Treatment standards for patients with red, yellow, and green labels was formulated, along with a unified diabetes medication catalog across the entire region. Through diabetes alliances and information platforms, doctors at regional medical center hospitals observed in real-time the blood glucose control and treatment status of patients at 10 secondary hospitals and 10 community hospitals, offering guidance to attain homogeneous and precise management(Fig.1,2,3,4). Following three years of implementation, this hierarchical medical system for diabetes encompassed 119,000 diabetes patients. The standardized management rate of diabetes at the primary healthcare level increased from 53.2% to 65.3%, the blood glucose control rate among patients rose from 55.8% to 62.7%, patient satisfaction improved from 94.2% to 98.63%. 87% of diabetes patients undergo routine management at the primary level, and the average inpatient cost per diabetes admission decreased by 7.3%. The volume of primary-level medical treatment increased by 15.16%. Compared with 2022, in 2025, the ratio of drugs and consumables decreased by 12.3%, and the proportion of medical service income increased by 7.2%. This study established a new model of hospital-wide blood glucose management with a “three-color label” grading, based on a high-quality dataset and artificial intelligence, and has achieved satisfactory outcomes, which is worthy of promotion and application.
