
Using AI to Predict Hospital Stays: BUV Lecturer’s Research Could Help Patients Recover Sooner and Ease Pressure on Hospitals
Jun 19, 2025
10:31:04
Dr. Dang Quang Vinh from the School of Computing & Innovative Technologies, together with his colleagues, have recently developed an AI-powered model that accurately predicts the length of hospital stay for patients undergoing surgery for adult spinal deformity. Their research paper “Predictive Modeling Of Length Of Stay in General Surgery Patients Using Artificial Intelligence”, was published in the Health Informatics Journal.
“How long should a patient stay in the hospital after a major surgery?” is a question that affects not just the patient, but also doctors, hospital staff, and healthcare systems trying to deliver quality care while managing limited resources. Dr. Dang Quang Vinh, a researcher and faculty member at BUV, is tackling this challenge with the help of Artificial Intelligence (AI).
In his latest study, Dr. Vinh focuses on patients undergoing surgery for adult spinal deformity (ASD) – a serious condition that often requires complex treatment and long recovery periods. His research introduces a machine learning model that can predict how long a patient is likely to stay in the hospital, based on their health data and surgical factors.
The study introduces a predictive model using the Gradient Ascent Decision Tree Model (GADTM) to estimate length of hospital stay (LOS). Through rigorous statistical techniques such as redundancy testing and univariable predictor screening, a clean and relevant set of variables was used to train the model.

The architecture of proposed method (GADTM)
Why is this important? Accurately predicting hospital stay has several real-world benefits:
- Patients can receive more personalised care: High-risk patients who may need extra support can be identified early, allowing for better planning and attention.
- Hospitals can operate more efficiently: By knowing in advance how long beds will be occupied, staff can plan discharges, admissions, and resources more effectively, therefore reducing overcrowding and delays.
- Costs can be reduced: Avoiding unnecessary hospital days helps both hospitals and patients save money.
- Recovery can be improved: Timely discharge and follow-up care help reduce complications and improve long-term outcomes.
Dr. Vinh’s model uses data from hundreds of patients and predicts LOS with high accuracy—within just two days of the actual stay.

GADTM shows significantly higher accuracy, precision and sensitivity than other existing methods
As AI tools like this continue to evolve, their use in hospitals could mean shorter stays, faster recoveries, and better use of public healthcare resources. Importantly, this is the first AI-based model specifically developed for ASD patients—a complex and underserved group for whom no specialised LOS prediction tool previously existed. Dr. Vinh’s work shows how advanced technology, when applied thoughtfully, can bring meaningful change to everyday healthcare—for both patients and society at large.
Read the full article here.
