Machine learning-based modeling of acute respiratory failure following emergency general surgery operations

Income-Based Disparities in Outcomes Following Pediatric Appendectomy: A National Analysis

BACKGROUND

Appendectomy remains a common pediatric surgical procedure with an estimated 80,000 operations performed each year. While prior work has reported the existence of racial disparities in postoperative outcomes, we sought to characterize potential income-based inequalities using a national cohort.

METHODS

All non-elective pediatric (<18 years) hospitalizations for appendectomy were tabulated in the 2016-2020 National Inpatient Sample. Only those in the highest (HI) and lowest income (LI) quartiles were considered for analysis. Multivariable regression models were developed to assess the independent association of income and postoperative major adverse events (MAE).

RESULTS

Of an estimated 87,830 patients, 36,845 (42.0%) were HI and 50,985 (58.0%) were LI. On average, LI patients were younger (11 [7-14] vs 12 [8-15] years, P < .001), more frequently insured by Medicaid (70.7 vs 27.3%, P < .05), and more commonly of Hispanic ethnicity (50.8 vs 23.4%, P < .001). Following risk adjustment, the LI cohort was associated with greater odds of MAE (adjusted odds ratio [AOR] 1.30 95% confidence interval [CI] 1.06-1.64). Specifically, low-income status was linked with increased odds of infectious (AOR 1.65, 95% CI 1.12-2.42) and respiratory (AOR 1.67, 95% CI 1.06-2.62) complications. Further, LI was associated with a $1670 decrement in costs ([2220-$1120]) and a +.32-day increase in duration of stay (95% CI [.21-.44]).

CONCLUSION

Pediatric patients of the lowest income quartile faced increased risk of major adverse events following appendectomy compared to those of highest income. Novel risk stratification methods and standardized care pathways are needed to ameliorate socioeconomic disparities in postoperative outcomes.

Machine Learning-based Prediction of Postoperative Pancreatic Fistula Following Pancreaticoduodenectomy

OBJECTIVE

The aim of this study was to develop a novel machine learning model to predict clinically relevant postoperative pancreatic fistula (CR-POPF) following pancreaticoduodenectomy (PD).

BACKGROUND

Accurate prognostication of CR-POPF may allow for risk stratification and adaptive treatment strategies for potential PD candidates. However, antecedent models, such as the modified Fistula Risk Score (mFRS), are limited by poor discrimination and calibration.

METHODS

All records entailing PD within the 2014 to 2018 American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) were identified. In addition, patients undergoing PD at our institution between 2013 and 2021 were queried from our local data repository. An eXtreme Gradient Boosting (XGBoost) model was developed to estimate the risk of CR-POPF using data from the ACS NSQIP and evaluated using institutional data. Model discrimination was estimated using the area under the receiver operating characteristic (AUROC) and area under the precision recall curve (AUPRC).

RESULTS

Overall, 12,281 and 445 patients undergoing PD were identified within the 2014 to 2018 ACS NSQIP and our institutional registry, respectively. Application of the XGBoost and mFRS scores to the internal validation dataset revealed that the former model had significantly greater AUROC (0.72 vs 0.68, P <0.001) and AUPRC (0.22 vs 0.18, P <0.001). Within the external validation dataset, the XGBoost model remained superior to the mFRS with an AUROC of 0.79 (95% CI: 0.74-0.84) versus 0.75 (95% CI: 0.70-0.80, P <0.001). In addition, AUPRC was higher for the XGBoost model, compared with the mFRS.

CONCLUSION

Our novel machine learning model consistently outperformed the previously validated mFRS within internal and external validation cohorts, thereby demonstrating its generalizability and utility for enhancing prediction of CR-POPF.

Machine learning improves prediction of postoperative outcomes after gastrointestinal surgery: a systematic review and meta-analysis

BACKGROUND

Machine learning (ML) approaches have become increasingly popular in predicting surgical outcomes. However, it is unknown whether they are superior to traditional statistical methods such as logistic regression (LR). This study aimed to perform a systematic review and meta-analysis to compare the performance of ML vs LR models in predicting postoperative outcomes for patients undergoing gastrointestinal (GI) surgery.

METHODS

A systematic search of Embase, MEDLINE, Cochrane, Web of Science, and Google Scholar was performed through December 2022. The primary outcome was the discriminatory performance of ML vs LR models as measured by the area under the receiver operating characteristic curve (AUC). A meta-analysis was then performed using a random effects model.

RESULTS

A total of 62 LR models and 143 ML models were included across 38 studies. On average, the best-performing ML models had a significantly higher AUC than the LR models (ΔAUC, 0.07; 95% CI, 0.04-0.09; P < .001). Similarly, on average, the best-performing ML models had a significantly higher logit (AUC) than the LR models (Δlogit [AUC], 0.41; 95% CI, 0.23-0.58; P < .001). Approximately half of studies (44%) were found to have a low risk of bias. Upon a subset analysis of only low-risk studies, the difference in logit (AUC) remained significant (ML vs LR, Δlogit [AUC], 0.40; 95% CI, 0.14-0.66; P = .009).

CONCLUSION

We found a significant improvement in discriminatory ability when using ML over LR algorithms in predicting postoperative outcomes for patients undergoing GI surgery. Subsequent efforts should establish standardized protocols for both developing and reporting studies using ML models and explore the practical implementation of these models.

Clinical and financial outcomes of transplant recipients following emergency general surgery operations

Introduction

Due to immunosuppression and underlying comorbidities, transplant recipients represent a vulnerable population following emergency general surgery (EGS) operations. The present study sought to evaluate clinical and financial outcomes of transplant patients undergoing EGS.

Methods

The 2010-2020 Nationwide Readmissions Database was queried for adults (≥18 years) with non-elective EGS. Operations included bowel resection, perforated ulcer repair, cholecystectomy, appendectomy and lysis of adhesions. Patients were classified by transplant history (Non-transplant, Kidney/Pancreas, Liver, Heart/Lung). The primary outcome was in-hospital mortality while perioperative complications, resource utilization and readmissions were secondarily considered. Multivariable regression models evaluated the association of transplant status on outcomes. Entropy balancing was employed to obtain a weighted comparison to adjust for intergroup differences.

Results

Of 7,914,815 patients undergoing EGS, 25,278 (0.32 %) had prior transplantation. The incidence of transplant patients increased temporally (2010: 0.23 %, 2020: 0.36 %, p < 0.001) with Kidney/Pancreas comprising the largest proportion (63.5 %). Non-transplant more frequently underwent appendectomy and cholecystectomy while transplant patients more commonly received bowel resections. Following entropy balancing, Liver was associated with decreased odds of mortality (AOR: 0.67, 95 % CI: 0.54-0.83, Reference: Non-transplant). Incremental hospitalization duration was longer in Liver and Heart/Lung compared to Non-transplant. Odds of acute kidney injury, readmissions and costs were higher in all transplant types.

Conclusion

The incidence of transplant recipients undergoing EGS operations has increased. Liver was observed to have lower mortality compared to Non-transplant. Transplant recipient status, regardless of organ, was associated with greater resource utilization and non-elective readmissions. Multidisciplinary care coordination is warranted to mitigate outcomes in this high-risk population.

A Surgeon's Guide to Artificial Intelligence-Driven Predictive Models

Artificial intelligence (AI) focuses on processing and interpreting complex information as well as identifying relationships and patterns among complex data. Artificial intelligence- and machine learning (ML)-driven predictions have shown promising potential in influencing real-time decisions and improving surgical outcomes by facilitating screening, diagnosis, risk assessment, preoperative planning, and shared decision-making. Fundamental understanding of the algorithms, as well as their development and interpretation, is essential for the evolution of AI in surgery. In this article, we provide surgeons with a fundamental understanding of AI-driven predictive models through an overview of common ML and deep learning algorithms, model development, performance metrics and interpretation. This would serve as a basis for understanding ML-based research, while fostering new ideas and innovations for furthering the reach of this emerging discipline.

ICD-10 based machine learning models outperform the Trauma and Injury Severity Score (TRISS) in survival prediction

Background

Precise models are necessary to estimate mortality risk following traumatic injury to inform clinical decision making or quantify hospital performance. The Trauma and Injury Severity Score (TRISS) has been the historical gold standard in survival prediction but its limitations are well-characterized. The present study used International Classification of Diseases 10^thRevision (ICD-10) injury codes with machine learning approaches to develop models whose performance was compared to that of TRISS.

Methods

The 2015–2017 National Trauma Data Bank was used to identify patients following trauma-related admission. Injury codes from ICD-10 were grouped by clinical relevance into 1,495 variables. The TRISS score, which comprises the Injury Severity Score, age, mechanism (blunt vs penetrating) as well as highest 24-hour values for systolic blood pressure (SBP), respiratory rate (RR) and Glasgow Coma Scale (GCS) was calculated for each patient. A base eXtreme gradient boosting model (XGBoost), a machine learning technique, was developed using injury variables as well as age, SBP, RR, mechanism and GCS. Prediction of in-hospital survival and other in-hospital complications were compared between both models using receiver operating characteristic (ROC) and reliability plots. A complete XGBoost model, containing injury variables, vitals, demographic information and comorbidities, was additionally developed.

Results

Of 1,380,740 patients, 1,338,417 (96.9%) survived to discharge. Compared to survivors, those who died were older and had a greater prevalence of penetrating injuries (18.0% vs 9.44%). The base XGBoost model demonstrated a greater receiver-operating characteristic (ROC) than TRISS (0.950 vs 0.907) which persisted across sub-populations and secondary endpoints. Furthermore, it exhibited high calibration across all risk levels (R² = 0.998 vs 0.816). The complete XGBoost model had an exceptional ROC of 0.960.

Conclusions

We report improved performance of machine learning models over TRISS. Our model may improve stratification of injury severity in clinical and quality improvement settings.