Publications by authors named "Ho-Taek Joo"
Commun Med (Lond)
November 2024
Article Synopsis
- Sepsis is a serious medical condition, and despite numerous clinical trials, finding effective treatment strategies is difficult due to small-scale testing; this study aims to extract optimal treatments from existing patient records.
- Using a modified deep reinforcement learning algorithm, researchers developed an AI model trained on over 16,000 hospital admissions, with performance tested through various statistical analyses and visualizations.
- The model showed a significant increase in estimated survival rates (up to 10.03%) and exhibited different treatment strategies compared to physicians, highlighting key factors like blood urea nitrogen and age, and indicating that while promising, results may need further clinical testing before implementation.
