PREDICTING PATIENT LENGTH OF STAY IN HEALTHCARE SETTINGS: A TWO-STAGE MACHINE LEARNING APPROACH UTILIZING ELECTRONIC RECORDS

Authors:

Dr. Patlannagari Hasitha Reddy, Dr. Persis Urbana Ivy B, Kayithi Kalpana

Page No: 637-651

Abstract:

The need for healthcare is increasing in Australia and globally. The healthcare system in Australia consists of a combination of commercial and governmental entities, including hospitals, clinics, and aged care institutions. The Australian healthcare system is notably economical and accessible, as around 68% of its funding is provided by the Australian government. In 2015-16, healthcare expenditure amounted to AUD 170.4 billion, or 10.0% of GDP. Escalating healthcare expenses and rising demand for services are intensifying the strain on the viability of the government-funded healthcare system. To achieve sustainability, we must enhance the efficiency of healthcare service delivery. If the demand for services is accurately understood, we can appropriately arrange the care delivery process and so enhance system efficiency. Nonetheless, the unpredictability of service demand contributes to inefficiencies in the healthcare delivery system. Escalating healthcare expenses and increasing service demand necessitate the more effective utilization of healthcare resources. The unpredictability of resource requirements diminishes the efficiency of the care delivery process. Our purpose is to diminish the ambiguity around patients' resource needs, which we do by categorizing individuals into analogous resource utilization categories. The traditional random forest and k-nearest neighbors (KNN) approaches yielded subpar classification and prediction results.

Description:

.

Volume & Issue

Volume-12,ISSUE-2

Keywords

Healthcare, K-Nearest Neighbors, Random Forest