Evaluating the Development of a Machine Learning Model for Predicting Length of Stay for Inpatients in a Tertiary...

Area of Science:

• Healthcare Informatics
• Predictive Analytics
• Clinical Decision Support

Background:

• Optimizing hospital resource allocation is crucial for efficiency.
• Patient-centered care and nursing workflow require effective management.
• Predictive modeling can enhance healthcare system performance.

Purpose of the Study:

• Develop a machine learning model to predict hospital length of stay.
• Incorporate clinical, nursing, and healthcare system factors.
• Improve resource allocation, patient care, and nursing efficiency.

Main Methods:

• Retrospective analysis of inpatient electronic medical records.
• Development and evaluation of machine learning algorithms.
• Identification of significant predictive variables using feature importance.

Main Results:

• Random Forest algorithm achieved highest accuracy in predicting length of stay.
• Key predictors include consultations, ICU stay, ventilator use, and infection isolation.
• Nursing factors like fall risk and pressure ulcer risk correlate with longer stays.

Conclusions:

• Machine learning models effectively predict hospital length of stay.
• Supports hospital resource management, nursing allocation, and patient safety.
• Predictive analytics enhance risk assessment, discharge planning, and hospital efficiency.