To effectively manage patients of different vulnerabilities to falls and pressure injury entails understanding the risk drivers and predicting risk profiles in real-time, thus, we determined the core drivers of patients’ proneness to these risks while developing a machine learning strategy for their real-time prediction in acute care hospital. By implementing a multivariate logistic analysis, the risk drivers and injury risk probabilities were obtained while establishing a comparative machine learning technique for patients’ risk-profiling. We observed Multi sclerosis & motor neuron disease (MSN) and Fall during current admission (FDA) as pronounced risk drivers, and Extra Tree Classifier (ETC) and Random Forest (RF) as the best algorithms with prediction accuracy of 90.6% - 99.8%. With a cost saving of 2.3% - 38.89%, our framework will provide an efficient technique for cost-effective management of inpatients susceptible to falls and pressure injury risks on admission.
I. OSSAI, Chinedu; O'Connor, Louise; and Wickramasinghe, Nilmini, "Real-time Inpatients Risk Profiling in Acute Care: A Comparative Study of Falls and Pressure Injuries Vulnerabilities" (2020). BLED 2020 Proceedings. 40.