A retrospective study of mortality for perioperative cardiac arrests toward a personalized treatment

Abstract

Perioperative cardiac arrest (POCA) is associated with a high mortality rate. This work aimed to study its prognostic factors for risk mitigation by means of care management and planning. A database of 380,919 surgeries was reviewed, and 150 POCAs were curated. The main outcome was mortality prior to hospital discharge. Patient demographic, medical history, and clinical characteristics (anesthesia and surgery) were the main features. Six machine learning (ML) algorithms, including LR, SVC, RF, GBM, AdaBoost, and VotingClassifier, were explored. The last algorithm was an ensemble of the first five algorithms. k-fold cross-validation and bootstrapping minimized the prediction bias and variance, respectively. Explainers (SHAP and LIME) were used to interpret the predictions. The ensemble provided the most accurate and robust predictions (AUC = 0.90 [95% CI, 0.78–0.98]) across various age groups. The risk factors were identified by order of importance. Surprisingly, the comorbidity of hypertension was found to have a protective effect on survival, which was reported by a recent study for the first time to our knowledge. The validated ensemble classifier in aid of the explainers improved the predictive differentiation, thereby deepening our understanding of POCA prognostication. It offers a holistic model-based approach for personalized anesthesia and surgical treatment.

Artificial Intelligence–enabled Decision Support in Surgery

“…The average AUROC for the best model across all 36 articles was 0.83; of the 8 articles with sample sizes of less than 2000 (ie, less than 1000 samples per class), 7 22,[26][27][28]31,35,37 (87.5%) had below-average (ie, less than 0.83) AUROC or accuracy. Twenty-four 9,18,20,[24][25][26][27][28][29][32][33][34][35]37,[39][40][41][43][44][45][46][47][48][49] articles (66.7%) presented confidence intervals for performance metrics. Among the 8 articles with sample sizes of less than 2000, 5 [26][27][28]35,37 presented confidence intervals.…”

“…Twentythree articles 9,19,21,[23][24][25]27,29,30,[32][33][34]36,37,39,41,42,[44][45][46][48][49][50] (63.9%) reported precision metrics (area under the precision-recall curve, positive predictive value, or F1 score). Twenty-five articles 9,20,21,[23][24][25][26][27][28]31,33,34,36,38,40,[42][43][44][45][46][47][48][49][50] (69.4%) included explainability mechanisms to convey the relative importance of input features in determining outputs. Thirteen articles 9,16,17,…”

“…Twenty-three articles9,19,21,23–25,27,29,30,32–34,36,37,39,41,42,44–46,48–50 (63.9%) reported precision metrics (area under the precision-recall curve, positive predictive value, or F1 score). Twenty-five articles9,20,21,23–28,31,33,34,36,38,40,42–50 (69.4%) included explainability mechanisms to convey the relative importance of input features in determining outputs. Thirteen articles9,16,17,20,25,27,29,30,35,38,45,46,50 (36.1%) presented a framework that could be used for clinical implementation; none of the articles assessed the efficacy of clinical implementation.…”

“…Twenty-five articles9,20,21,23–28,31,33,34,36,38,40,42–50 (69.4%) included explainability mechanisms to convey the relative importance of input features in determining outputs. Thirteen articles9,16,17,20,25,27,29,30,35,38,45,46,50 (36.1%) presented a framework that could be used for clinical implementation; none of the articles assessed the efficacy of clinical implementation.…”

“…Twenty-five articles 9,20,21,[23][24][25][26][27][28]31,33,34,36,38,40,[42][43][44][45][46][47][48][49][50] (69.4%) included explainability mechanisms to convey the relative importance of input features in determining outputs. Thirteen articles 9,16,17,20,25,27,29,30,35,38,45,46,50 (36.1%) presented a framework that could be used for clinical implementation; none of the articles assessed the efficacy of clinical implementation. in 5 articles performed external or real-time validation; 1 in 5 likely might have had overfitting and lower predictive performance due to small sample sizes; only two-thirds presented confidence intervals for performance metrics; less than two-thirds reported measures of precision; and none of the articles reported analyses across sex or racial groups or assessed the efficacy of clinical implementation.…”