Patient Pathway Optimization in Hospitals using Convolutional Neural Network and m-Artificial Bee Colony Algorithm

Patients, either scheduled or unscheduled, have to be put on pathways depending on their clinical conditions. These pathways when assigned to patients, at times result in prolonged waiting times. The patient pathway has to be selected in such a way that their waiting times is minimized. To accomplish this, real-time optimization of workflow is required. In this study, we propose 2-fold solution, first to use image processing techniques to capture system status of all departments in the hospital. A Convolutional Neural Network is trained to recognize the patients. The number of patients waiting for the care in each department is identified and the database is updated in real time. Second, to use this real-time system status, to select optimal pathways using modified Artificial Bee Colony (m-ABC) algorithm. Image processing technique captures and provides more accurate system status for real-time workflow optimization. The ABC algorithm is modified by incorporating the precedence constraints that clinical workflows have. By selecting optimal pathway through m-ABC, the waiting times of patients is minimized.