Real-Time Predictive Analytics in Healthcare Using Deep Q-Networks

The purpose of this research is to utilize Deep Q-Networks (DQNs) for real time predictive analytics in health care to promote improved patient’s well-being and efficiency of resources utilization. Based on the EHRs and monitoring data in real time, the effectiveness of the proposed methodology lies in the method of using the DQNs for early identification of the potential patient deterioration, early onset of diseases, admission, and discharge from the premises of the hospital. The DQN model was also plausible in terms of keeping the accuracy of predicting patient deterioration at 85% and recalled 80% of the patients to diagnose sepsis early. Further, accuracy of hospital admission and discharge was found to be 90% and efficiency of utilization of hospital resources and the number of new patients was found to be 92% in the proposed model. Therefore, the significance of data quality for DQNs in healthcare delivery is underscored, in addition to issues of data privacy, model interpretability, and other ethic concerns highlighted in this study for successful implementation.