November 1, 2022Journal of Clinical MedicineOpen Access

Prediction Model for 30-Day Mortality after Non-Cardiac Surgery Using Machine-Learning Techniques Based on Preoperative Evaluation of Electronic Medical Records

Q: What is the clinical evidence from this study?

Study design: Observational. Population: Non-cardiac surgery (n=339725). Intervention: Extreme gradient boosting (XGB) prediction model vs. Baseline logistic regression model. Primary outcome: Prediction of 30-day mortality (AUROC in external validation) (AUROC 0.932, 95% CI 0.919-0.945).

View Full Paper

Ask AI

Bookmark

Key result

Machine-learning model using preoperative EMR data predicts 30-day mortality after non-cardiac surgery with ~0.93 AUROC.

AUROC 0.932
95% CI 0.919-0.945
n=339,725

Why the study?

Machine-learning techniques are useful for clinical prediction models, and the study aimed to construct a prediction model for postoperative 30-day mortality using automatically extracted electronic preoperative evaluation sheets.

Can a machine-learning prediction model based on preoperative electronic medical records accurately predict 30-day mortality after non-cardiac surgery?

Population

276,341 development/internal validation and 63,384 external validation adult non-cardiac surgery patients

Design

Prediction model development and multicenter validation study

Follow-up

30 days

Authors

BCByungjin Choi AOAh Ran Oh SLSeung Hwa Lee

Discussion

Member takes

Overview

May facilitate automated preoperative risk stratification; extends ML models but leaves open prospective validation before clinical adoption.

Study Design

Type

Observational (n=339,725)

Multicenter

Yes

Structured PICO

Can a machine-learning prediction model based on preoperative electronic medical records accurately predict 30-day mortality after non-cardiac surgery?

Population

339,725 adult patients undergoing non-cardiac surgery across two centers in South Korea, evaluated for 30-day postoperative mortality.

Exposure

Extreme gradient boosting (XGB) machine-learning prediction model based on automatically extracted electronic preoperative evaluation sheets

Outcome

Postoperative 30-day mortalityhard clinical

Main Result

Effect estimate: AUROC 0.932 (95% CI 0.919-0.945)

Absolute Event Rate: 0.932% vs 0.723%

A machine-learning model using automatically extracted preoperative EMR data can accurately predict 30-day mortality after non-cardiac surgery.

Limitations

Retrospective data design means causality cannot be confirmed.
Included patients regardless of surgery types, which may limit clinical meaningfulness for palliative or life-critical emergency surgeries.
Cannot be fully generalized, especially for other ethnicities.
Perioperative care may not have been controlled and institutional protocols may vary.
Imbalanced dataset with a low incidence of the primary outcome.
Cannot confirm whether factor modification could improve outcomes or if model adoption would be clinically helpful.

Cite This Study

Choi et al. (2022) conducted an observational in Non-cardiac surgery (n=339,725). Extreme gradient boosting (XGB) prediction model vs. Baseline logistic regression model was evaluated on Prediction of 30-day mortality (AUROC in external validation) (AUROC 0.932, 95% CI 0.919-0.945). An extreme gradient boosting machine-learning model using automatically extracted preoperative electronic medical record data accurately predicted 30-day mortality after non-cardiac surgery, achieving an AUROC of 0.932 in external validation.

synapsesocial.com/papers/6aa499edfc0ec4e314f2d00c https://doi.org/10.3390/jcm11216487

View Full Paper

Ask AI

Bookmark

November 1, 2022Journal of Clinical MedicineOpen Access

Prediction Model for 30-Day Mortality after Non-Cardiac Surgery Using Machine-Learning Techniques Based on Preoperative Evaluation of Electronic Medical Records

View Full Paper

Ask AI

Bookmark

Key result

Machine-learning model using preoperative EMR data predicts 30-day mortality after non-cardiac surgery with ~0.93 AUROC.

AUROC 0.932
95% CI 0.919-0.945
n=339,725

Why the study?

Can a machine-learning prediction model based on preoperative electronic medical records accurately predict 30-day mortality after non-cardiac surgery?

Population

276,341 development/internal validation and 63,384 external validation adult non-cardiac surgery patients

Design

Prediction model development and multicenter validation study

Follow-up

30 days

Authors

BCByungjin Choi AOAh Ran Oh SLSeung Hwa Lee

Discussion

Member takes

Overview

May facilitate automated preoperative risk stratification; extends ML models but leaves open prospective validation before clinical adoption.

Study Design

Type

Observational (n=339,725)

Multicenter

Yes

Structured PICO

Can a machine-learning prediction model based on preoperative electronic medical records accurately predict 30-day mortality after non-cardiac surgery?

Population

339,725 adult patients undergoing non-cardiac surgery across two centers in South Korea, evaluated for 30-day postoperative mortality.

Exposure

Extreme gradient boosting (XGB) machine-learning prediction model based on automatically extracted electronic preoperative evaluation sheets

Outcome

Postoperative 30-day mortalityhard clinical

Main Result

Effect estimate: AUROC 0.932 (95% CI 0.919-0.945)

Absolute Event Rate: 0.932% vs 0.723%

A machine-learning model using automatically extracted preoperative EMR data can accurately predict 30-day mortality after non-cardiac surgery.

Limitations

Retrospective data design means causality cannot be confirmed.
Included patients regardless of surgery types, which may limit clinical meaningfulness for palliative or life-critical emergency surgeries.
Cannot be fully generalized, especially for other ethnicities.
Perioperative care may not have been controlled and institutional protocols may vary.
Imbalanced dataset with a low incidence of the primary outcome.
Cannot confirm whether factor modification could improve outcomes or if model adoption would be clinically helpful.

Cite This Study

synapsesocial.com/papers/6aa499edfc0ec4e314f2d00c https://doi.org/10.3390/jcm11216487

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark