Development of postoperative delirium prediction models in patients undergoing cardiovascular surgery using machine learning algorithms
Problem
Postoperative delirium after cardiovascular surgery is common and linked to adverse events; high-risk patients need preoperative identification.
Why it matters
Early identification enables focused interventions to prevent delirium-related adverse events.
What the AI did
Machine learning models using preoperative and intraoperative factors were developed to predict delirium, validated via stratified fivefold cross-validation.
Results
Cohort: 87 patients, 24 (27.6%) developed delirium; model performance not stated in abstract
Impact
Prospective single-hospital study developed prediction models; performance metrics not shown in abstract.
Record
| Venue | Scientific Reports |
|---|---|
| Published | 2023 |
| Work type | article |
| Authors | Chie Nagata, Masahiro Hata, Yuki Miyazaki, Hirotada Masuda, Tamiki Wada, Tasuku Kimura, Makoto Fujii, Yasushi Sakurai et al. |
| Citations | 30 (OpenAlex, at import time) |
| References | 63 |
| Model or method | Bernoulli naive Bayes, SVM, Random forest, Extra-trees, XGBoost |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41598-023-48418-5 |
| Open access | gold |
| Keywords | delirium, cardiovascular surgery, machine learning, XGBoost, prediction |
Importance rating
real but narrower advance
Delirium is linked to postoperative adverse events; single-hospital prospective study of 87 patients with modest model performance (AUROC 0.76).