Related Experiment Videos
FT-Transformer DeepSurv: A Novel Approach to Predicting Post-Discharge Mortality in Older Adults.
Jae Yeon Chung1, Sung Hwan Ji1, Jun Sang Yoo1
1Department of Digital Health, SAIHST, Sungkyunkwan University, Seoul, Korea.
Studies in Health Technology and Informatics
|July 3, 2026
Summary
A new AI model accurately predicts mortality risk for elderly emergency department patients, aiding advance directive discussions. It identifies inflammation and comorbidities as key factors, outperforming age alone.
Area of Science:
- Gerontology
- Artificial Intelligence in Medicine
- Health Informatics
Background:
- Validated tools for predicting mortality in older emergency department (ED) patients (≥65) are lacking.
- This gap hinders timely Advance Directive (AD) engagement for end-of-life care planning.
Purpose of the Study:
- To develop and validate an explainable machine learning model for predicting hospital discharge mortality in elderly ED patients.
- To identify key predictors of mortality beyond chronological age.
Main Methods:
- Utilized a large dataset of 41,612 electronic health records (EHR) from Samsung Medical Center (2018-2022).
- Developed and compared four predictive models, including a FT-Transformer DeepSurv model.
- Evaluated model performance using the C-Index and generalization gap.
Main Results:
- The FT-Transformer DeepSurv model achieved a high test C-Index of 0.8747.
- This model demonstrated the smallest generalization gap (0.015) among the compared models.
- Inflammation and comorbidity burden were identified as more significant predictors of mortality than age.
Conclusions:
- An explainable AI model can accurately predict discharge mortality in elderly ED patients.
- This tool can support safer discharge planning and facilitate proactive AD discussions for high-risk individuals.
- The model highlights the importance of biological factors (inflammation, comorbidities) over age in mortality prediction for this demographic.
Related Concept Videos
01:20Actuarial Approach
The actuarial approach, a statistical method originally developed for life insurance risk assessment, is widely used to calculate survival rates in clinical and population studies. This method accounts for participants lost to follow-up or those who die from causes unrelated to the study, ensuring a more accurate representation of survival probabilities.
Consider the example of a high-risk surgical procedure with significant early-stage mortality. A two-year clinical study is conducted,...
Consider the example of a high-risk surgical procedure with significant early-stage mortality. A two-year clinical study is conducted,...
01:21Energy Losses in Transformers
In an ideal transformer, it is assumed that there are no energy losses, and, hence, all the power at the primary winding is transferred to the secondary winding. However, in reality, the transformers always have some energy losses, and, hence, the output power obtained at the secondary winding is less than the input power at the primary winding due to energy losses.
There are four main reasons for energy losses in transformers.
The first cause can be the high resistance of the copper windings...
There are four main reasons for energy losses in transformers.
The first cause can be the high resistance of the copper windings...