Determining Factors Influencing Length of Stay and Predicting Length of Stay Using Data Mining in the General Surgery Department

Length of stay is one of the most important indicators in assessing hospital performance. A shorter stay can reduce the costs per discharge and shift care from inpatient to less expensive post-acute settings. It can lead to a greater readmission rate, better resource management, and more efficient services.
This study aimed to identify the factors influencing length of hospital stay and predict length of stay in the general surgery department.
In this study, patient information was collected from 327 records in the surgery department of Shariati Hospital using data mining techniques to determine factors influencing length of stay and to predict length of stay using three algorithms, namely decision tree, Naïve Bayes, and k-nearest neighbor algorithms. The data was split into a training data set and a test data set, and a model was built for the training data. A confusion matrix was obtained to calculate accuracy.
Four factors presented: surgery type (hemorrhoid), average number of visits per day, number of trials, and number of days of hospitalization before surgery; the most important of these factors was length of stay. The overall accuracy of the decision tree was 88.9% for the training data set.
This study determined that all three algorithms can predict length of stay, but the decision tree performs the best.