This project seeks to leverage Machine Learning to provide useful predictions of the general Bed Occupancy Rates (BOR) in Singapore public hospitals. BOR forecasting is useful to mitigate the ongoing problem of insufficient beds and long wait time for hospital admissions, facilitating bed capacity management. We found that Support Vector Classifier outperformed all ML models implemented in our study, and is a suitable model for our BOR prediction problem.
The datasets used to train and test the models are manually curated and contains data pulled from publicly available databases or documents. In total, 5 models are trained and tested for its effectiveness in predicting hospital bed occupancy rates. The models are Decision Tree, Naive Bayes Classifier, Support Vector Classifier, K-Nearest Neighbours and Multi-Layer Perceptron.
Code used to train, optimise, test and evaluate the 5 models are implemented on Google Colab.
This project is a joint effort between Amanda Ho, Amos Chan, Daniel Chan, Hemanshu Ghandi, Matthew Lee and Tiana Chen.