A sensing-based patient classification framework for efficient patient-nurse scheduling

Nowadays, hospitals and government departments are struggling to reduce the health costs and improve the service quality. Hospitals rely mainly on nurses who have many duties including caring of patients, communicating with doctors, administering medicine and checking vital signs. Wireless body sensor network (WBSN) is one of the main applications that helps the medical staff in electronic health surveillance with early detection of critical physiological symptoms. However, the rapid growth in the patient number poses a high pressure on hospitals and increases the nurse duties. In addition, WBSN suffers from big data collected by the medical sensors. Therefore, collecting data is one of the major reason behind the exhaustion of limited energy. In this paper, we propose a novel and efficient framework for nurse-patient intelligent task organization that treats the scheduling problem in hospitals. Typically, our framework consists of three phases: data collection and transmission, patient classification, and nurse scheduling. The first phase uses the early warning system (EWS) to update the medical staff when the patient situation is changed. This will help to minimize the energy consumption in sensor and control its data transmission rate. Once the patient data are collected, the second level introduces an efficient classification model that grouping patients according to the severity level of their vital signs. The last phase proposes a nurse-scheduling algorithm based on the classification results and some priority metrics to balance the workload of the stressed medical personnel. We performed our simulation on real health sensor data collected from 18 patients, including various vital signs, while assuming a certain number of nurses. The obtained results show the effectiveness of our framework in conserving the sensor energy up to 95%, and ensuring a balanced nurse workloads where each nurse serves, during each period of one hour, an average of 5 patients with corresponding cumulative criticality of 62%.