Hybrid feature selection in a machine learning predictive model for perioperative myocardial injury in noncoronary cardiac surgery with cardiopulmonary bypass

BACKGROUND

Perioperative myocardial injury (PMI) is associated with increased mobility and mortality after noncoronary cardiac surgery. However, limited studies have developed a predictive model for PMI. Therefore, we used hybrid feature selection (FS) methods to establish a predictive model for PMI in noncoronary cardiac surgery with cardiopulmonary bypass (CPB).

METHODS

This was a single-center retrospective study conducted at the Fuwai Hospital in China. Patients aged 18-70 years who underwent elective noncoronary surgery with CPB at our institution from December 2018 to April 2021 were enrolled. The primary outcome was PMI, defined as the postoperative cardiac troponin I (cTnI) levels exceeding 220 times of upper reference limit (URL). Statistical analyses were conducted by Python (Python Software Foundation, version 3.9.7 and integrated development environment Jupyter Notebook 1.1.0) and SPSS software version 26.0 (IBM Corp., Armonk, New York, USA).

RESULTS

A total of 1130 patients were eventually eligible for this study. The incidence of PMI was 20.3% (229/1130) in the overall patients, 20.6% (163/791) in the training dataset, and 19.5% (66/339) in the testing dataset. The logistic regression model performed the best AUC of 0.6893 (95 CI%: 0.6371-0.7382) by the traditional selection method, and the random forest model performed the best AUC of 0.6937 (95 CI%: 0.6416-0.7423) by the union of Wrapper and Embedded method, and the CatBoost model performed the best AUC of 0.6828 (95 CI%: 0.6304-0.7320) by the union of Embedded and forward logistic regression technique, and the Naïve Bayes model achieved the best AUC with 0.7254 (95 CI%: 0.6746-0.7723) by forwarding logistic regression method. Moreover, the decision tree, KNeighborsClassifier, and support vector machine models performed the worse AUC in all selection forms. Furthermore, the SHapley Additive exPlanations plot showed that prolonged CPB, aortic clamp time, and preoperative low platelets count were strongly related to the PMI risk.

CONCLUSIONS

In total, four category feature selection methods were utilized, comprising five individual selection techniques and 15 combined methods. Notably, the combination of logistic regression and embedded methods demonstrated outstanding performance in predicting PMI risk. We also concluded that the machine learning model, including random forest, catboost, and Naive Bayes, were suitable candidates for establishing PMI predictive model. Nevertheless, additional investigation and validation are imperative for substantiating these finding.

Incidence and risk factors of postoperative acute myocardial injury in noncardiac patients: A systematic review and meta-analysis

INTRODUCTION

Postoperative myocardial injury after noncardiac surgery is common and is associated with short- and long-term morbidity and mortality. However, the incidence and risk factors for postoperative acute myocardial injury (POAMI) are currently unknown due to inconsistent definitions.

METHODS

We systematically searched PubMed and Web of Science to identify studies that applied the change value of preoperative and postoperative cardiac troponins to define cardiac injury. We estimated the pooled incidence, risk factors, and 30-day and long-term mortality of POAMI in noncardiac patients. The study protocol was registered with PROSPERO, CRD42023401607.

RESULTS

Ten cohorts containing 11,494 patients were included for analysis. The pooled incidence of POAMI was 20% (95% CI: 16% to 23%). Preoperative hypertension (OR: 1.47; 95% CI: 1.30 to 1.66), cardiac failure (OR: 2.63; 95% CI: 2.01 to 3.44), renal impairment (OR: 1.66; 95% CI: 1.48 to 1.86), diabetes (OR: 1.43; 95% CI: 1.27 to 1.61), and preoperative beta-blocker intake (OR: 1.65; 95% CI: 1.10 to 2.49) were the risk factors for POAMI. Age (mean difference: 2.08 years; 95% CI: -0.47 to 4.62), sex (male, OR: 1.16; 95% CI: 0.77 to 1.76), body mass index (mean difference: 0.35; 95% CI: -0.86 to 1.57), preoperative coronary artery disease (OR: 2.10; 95% CI: 0.85 to 5.21), stroke (OR: 0.90; 95% CI: 0.50 to 1.59) and preoperative statins intake (OR: 0.65; 95% CI: 0.21 to 2.02) were not associated with POAMI. Patients with POAMI had higher preoperative hsTnT levels (mean difference: 5.92 ng/L; 95% CI: 4.17 to 7.67) and lower preoperative hemoglobin levels (mean difference: -1.29 g/dL; 95% CI: -1.43 to -1.15) than patients without.

CONCLUSION

Based on this meta-analysis, approximately 1 in 5 of noncardiac patients develop POAMI. However, the lack of a universally recognized definition for POAMI, which incorporates diverse cardiac biomarkers and patient groups, poses a challenge in accurately characterizing its incidence, risk factors, and clinical outcomes.

Association between preoperative evaluation with lung ultrasound and outcome in frail elderly patients undergoing orthopedic surgery for hip fractures: study protocol for an Italian multicenter observational prospective study (LUSHIP)

BACKGROUND

Hip fracture is one of the most common orthopedic causes of hospital admission in frail elderly patients. Hip fracture fixation in this class of patients is considered a high-risk procedure. Preoperative physical examination, plasma natriuretic peptide levels (BNP, Pro-BNP), and cardiovascular scoring systems (ASA-PS, RCRI, NSQIP-MICA) have all been demonstrated to underestimate the risk of postoperative complications. We designed a prospective multicenter observational study to assess whether preoperative lung ultrasound examination can predict better postoperative events thanks to the additional information they provide in the form of "indirect" and "direct" cardiac and pulmonary lung ultrasound signs.

METHODS

LUSHIP is an Italian multicenter prospective observational study. Patients will be recruited on a nation-wide scale in the 12 participating centers. Patients aged  >  65 years undergoing spinal anesthesia for hip fracture fixation will be enrolled. A lung ultrasound score (LUS) will be generated based on the examination of six areas of each lung and ascribing to each area one of the four recognized aeration patterns-each of which is assigned a subscore of 0, 1, 2, or 3. Thus, the total score will have the potential to range from a minimum of 0 to a maximum of 36. The association between 30-day postoperative complications of cardiac and/or pulmonary origin and the overall mortality will be studied. Considering the fact that cardiac complications in patients undergoing hip surgery occur in approx. 30% of cases, to achieve 80% statistical power, we will need a sample size of 877 patients considering a relative risk of 1.5.

CONCLUSIONS

Lung ultrasound (LU), as a tool within the anesthesiologist's armamentarium, is becoming increasingly widespread, and its use in the preoperative setting is also starting to become more common. Should the study demonstrate the ability of LU to predict postoperative cardiac and pulmonary complications in hip fracture patients, a randomized clinical trial will be designed with the scope of improving patient outcome. Trial registration ClinicalTrials.gov, NCT04074876. Registered on August 30, 2019.