Thumbnail Image
Publication

Domain Specific Transformers-Based Prioritization of Re-Admission for Patients in Healthcare

Citations
Altmetric:
Video URL
Date
2023
Author
Naseem, Hira, Catal, Cagatay
Collections
Computer Science & Engineering
Show all metadata
Files
Thumbnail Image
3639233.3639350.pdf
Adobe PDF, 2.01 MB
Abstract
Hospital readmissions, indicative of healthcare quality and resource-intensive, require improved efficiency through accurately identifying higher-risk patients; while numerous prediction models exist for re-Admissions, none have prioritized them specifically. In this research, we focused on creating a predictive model designed to prioritize patient readmissions based on their discharge summaries from MIMIC III. To improve the model's performance and address class imbalance problems in this domain, we applied data augmentation. BioBERT was employed for feature extraction during model training, and the Optuna framework was utilized for hyperparameter optimization. The predictive model demonstrated remarkable performance when evaluated on the testing dataset, achieving AUROC score of 0.69. The experimental results highlight the model's potential to effectively prioritize patient re-Admissions, thereby contributing to more informed and efficient healthcare decision-making processes.
URI
http://hdl.handle.net/10576/70181