Predictive Modeling of Readmission in Patients with Schizophrenia Using Machine Learning Models

Tennison Daniel, Daphne Shekinah (2024) Predictive Modeling of Readmission in Patients with Schizophrenia Using Machine Learning Models. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (784kB) | Preview

Abstract

Schizophrenia is a mental disorder which is chronic, and it affects a person’s ability to grasp reality. It not only affects the person who’s schizophrenic but also the people surrounding the said person. There are more than 20 million people who are affected by schizophrenia but not many people get the required treatment. This leads to hospitalization and re-hospitalization of patients. There are many rules and regulations set in place for the treatment and the discharge of a patient in different countries. Predicting the risk of a patient’s hospital readmission would help not only the patient but also the healthcare professionals who are treating the patient, because learning about how the disorder is affecting the person and treating them with a specific type of care could prove to be helpful. To address this issue, many machine learning models were developed. But there aren’t many papers or research done that was specific to Ireland. So, in this project, a dataset that was published by the National Psychiatric Inpatient Reporting System under the Health Research Board from the year 2006 to 2022 was used to predict the readmission of patients with schizophrenia and other related disorders. The proposed model, XGBoost classifier performed the best, even when there was quite a class imbalance. The model achieved an accuracy of 75.38%, recall of 81%, F1-Score of 76% and the precision of the model was 72%. If imputation had been used on the dataset instead of filling up the missing values, the accuracy of the model could’ve been increased by a significant amount. By using a dataset that does not have many missing values, and combining the model with SMOTE could improve the results drastically.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Tomer, Vikas UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science R Medicine > Healthcare Industry Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning R Medicine > RA Public aspects of medicine > RA790 Mental Health
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	05 Sep 2025 11:36
Last Modified:	05 Sep 2025 11:36
URI:	https://norma.ncirl.ie/id/eprint/8825

Actions (login required)

View Item