What question did this study set out to answer?

To develop and integrate a machine learning-based clinical tool to predict postoperative adverse outcomes, ICU admission, and prolonged hospital stay in patients undergoing hip repair surgery.

April 23, 2022BMC AnesthesiologyOpen Access

Implementation of a machine learning application in preoperative risk assessment for hip repair surgery

View Full Paper

Ask AI

Bookmark

Why the study?

The study aimed to develop a machine learning application to assist anesthesiologists in assessing complication risks in patients undergoing hip surgery.

Does a machine learning-based application improve preoperative risk assessment for adverse outcomes compared to ASA-PS in adult patients undergoing hip repair surgery?

Population

4,448 adult patients undergoing hip repair surgery

Comparison

Machine learning model vs ASA-PS

Design

Retrospective multicenter prediction model development and validation study

Authors

YLYuyu LiShandong University JWJhi‐Joung WangTri-Service General Hospital SHSheng‐Han HuangChang Gung Memorial Hospital

Discussion

Member takes

Implication

May enhance preoperative risk prediction in hip repair; leaves open prospective validation before clinical adoption.

Key Points

To develop and integrate a machine learning-based clinical tool to predict postoperative adverse outcomes, ICU admission, and prolonged hospital stay in patients undergoing hip repair surgery.
Analyzed electronic health record data from 4,448 adult patients undergoing hip repair surgery across three hospital centers between January 2013 and March 2020.
Trained and evaluated seven machine learning algorithms using 22 preoperative clinical features, utilizing a 70/30 train-test split.
Integrated the top-performing model into the hospital information system and compared its predictive power against the standard American Society of Anesthesiologists Physical Status (ASA-PS) classification.
Among 4,448 patients, 102 (2.3%) experienced primary composite adverse events, 160 (3.6%) required ICU admission, and 401 (9.0%) had prolonged length of stay.
The optimal machine learning model significantly outperformed ASA-PS in predicting composite adverse outcomes (AUROC 0.810 vs. 0.629, p < 0.01).
The model also achieved superior accuracy over ASA-PS for predicting ICU admission (AUROC 0.835 vs. 0.692, p < 0.01) and prolonged hospital stay (AUROC 0.832 vs. 0.618, p < 0.01).

Structured PICO

Does a machine learning-based application improve preoperative risk assessment for adverse outcomes compared to ASA-PS in adult patients undergoing hip repair surgery?

Population

4,448 adult patients aged 20 years and above who underwent surgical treatment of a hip fracture at Chi Mei Medical Center and its 2 branch hospitals in Tainan City, Taiwan, from January 1, 2013, to March 31, 2020. Mean age 65.3 years, 42.4% male.

Intervention

Machine learning-based application integrated into the hospital information system (HIS) for preoperative risk assessment

Comparator

American Society of Anesthesiologists Physical Status (ASA-PS) scoring

Outcome

Composite of postoperative adverse events, including in-hospital mortality (and death within 48 h after discharge), acute stroke, acute myocardial infarction, acute respiratory failure, sepsis, acute liver failure, and acute renal failurecomposite

A hospital-specific machine learning model outperformed traditional ASA-PS scoring in predicting postoperative adverse outcomes, ICU admission, and prolonged length of stay in patients undergoing hip repair surgery.

Cite This Study

Li et al. (2022) studied this question.

synapsesocial.com/papers/6a7f769bfd903b7756ca781d https://doi.org/10.1186/s12871-022-01648-y

View Full Paper

Ask AI

Bookmark

Implementation of a machine learning application in preoperative risk assessment for hip repair surgery

Discussion

Member takes

Implication

Key Points

Structured PICO

Cite This Study

Implementation of a machine learning application in preoperative risk assessment for hip repair surgery

Discussion

Member takes

Implication

Key Points

Structured PICO

Cite This Study

Also Consider

Also Consider