Utilizing Machine Learning Methods for Preoperative Prediction of Postsurgical Mortality and Intensive Care Unit Admission

Mitigating Sociodemographic Bias in Opioid Use Disorder Prediction: Fairness-Aware Machine Learning Framework

BACKGROUND

Opioid use disorder (OUD) is a critical public health crisis in the United States, affecting >5.5 million Americans in 2021. Machine learning has been used to predict patient risk of incident OUD. However, little is known about the fairness and bias of these predictive models.

OBJECTIVE

The aims of this study are two-fold: (1) to develop a machine learning bias mitigation algorithm for sociodemographic features and (2) to develop a fairness-aware weighted majority voting (WMV) classifier for OUD prediction.

METHODS

We used the 2020 National Survey on Drug and Health data to develop a neural network (NN) model using stochastic gradient descent (SGD; NN-SGD) and an NN model using Adam (NN-Adam) optimizers and evaluated sociodemographic bias by comparing the area under the curve values. A bias mitigation algorithm, based on equality of odds, was implemented to minimize disparities in specificity and recall. Finally, a WMV classifier was developed for fairness-aware prediction of OUD. To further analyze bias detection and mitigation, we did a 1-N matching of OUD to non-OUD cases, controlling for socioeconomic variables, and evaluated the performance of the proposed bias mitigation algorithm and WMV classifier.

RESULTS

Our bias mitigation algorithm substantially reduced bias with NN-SGD, by 21.66% for sex, 1.48% for race, and 21.04% for income, and with NN-Adam by 16.96% for sex, 8.87% for marital status, 8.45% for working condition, and 41.62% for race. The fairness-aware WMV classifier achieved a recall of 85.37% and 92.68% and an accuracy of 58.85% and 90.21% using NN-SGD and NN-Adam, respectively. The results after matching also indicated remarkable bias reduction with NN-SGD and NN-Adam, respectively, as follows: sex (0.14% vs 0.97%), marital status (12.95% vs 10.33%), working condition (14.79% vs 15.33%), race (60.13% vs 41.71%), and income (0.35% vs 2.21%). Moreover, the fairness-aware WMV classifier achieved high performance with a recall of 100% and 85.37% and an accuracy of 73.20% and 89.38% using NN-SGD and NN-Adam, respectively.

CONCLUSIONS

The application of the proposed bias mitigation algorithm shows promise in reducing sociodemographic bias, with the WMV classifier confirming bias reduction and high performance in OUD prediction.

Artificial Intelligence in Perioperative Care: Opportunities and Challenges

Artificial intelligence (AI) applications have great potential to enhance perioperative care. This paper explores promising areas for AI in anesthesiology; expertise, stakeholders, and infrastructure for development; and barriers and challenges to implementation.

Combination of Creatinine with Inflammatory Biomarkers (PCT, CRP, hsCRP) for Predicting Postoperative ICU Admissions for Elderly Patients

INTRODUCTION

The number of elderly patients who require surgery as their primary treatment has increased rapidly in recent years. Among 300 million people globally who underwent surgery every year, patients aged 65 years and over accounted for more than 30% of cases. Despite medical advances, older patients remain at higher risk of postoperative complications. Early diagnosis and effective prediction are essential requirements for preventing serious postoperative complications. In this study, we aim to provide new biomarker combinations to predict the incidence of postoperative intensive care unit (ICU) admissions > 24 h in elderly patients.

METHODS

This investigation was conducted as a nested case-control study, incorporating 413 participants aged ≥ 65 years who underwent non-cardiac, non-urological elective surgeries. These individuals underwent a 30-day postoperative follow-up. Before surgery, peripheral venous blood was collected for analyzing serum creatinine (Scr), procalcitonin (PCT), C-reactive protein (CRP), and high-sensitivity CRP (hsCRP). The efficacy of these biomarkers in predicting postoperative complications was evaluated using receiver operating characteristic (ROC) curve analysis and area under the curve (AUC) values.

RESULTS

Postoperatively, 10 patients (2.42%) required ICU admission. Regarding ICU admissions, the AUCs with 95% confidence intervals (CIs) for the biomarker combinations of Scr × PCT and Scr × CRP were 0.750 (0.655-0.845, P = 0.007) and 0.724 (0.567-0.882, P = 0.015), respectively. Furthermore, cardiovascular events were observed in 14 patients (3.39%). The AUC with a 95% CI for the combination of Scr × CRP in predicting cardiovascular events was 0.688 (0.560-0.817, P = 0.017).

CONCLUSION

The innovative combinations of biomarkers (Scr × PCT and Scr × CRP) demonstrated efficacy as predictors for postoperative ICU admissions in elderly patients. Additionally, the Scr × CRP also had a moderate predictive value for postoperative cardiovascular events.

TRIAL REGISTRATION

China Clinical Trial Registry, ChiCTR1900026223.

Explainable artificial intelligence prediction-based model in laparoscopic liver surgery for segments 7 and 8: an international multicenter study

BACKGROUND

Artificial intelligence (AI) is becoming more useful as a decision-making and outcomes predictor tool. We have developed AI models to predict surgical complexity and the postoperative course in laparoscopic liver surgery for segments 7 and 8.

METHODS

We included patients with lesions located in segments 7 and 8 operated by minimally invasive liver surgery from an international multi-institutional database. We have employed AI models to predict surgical complexity and postoperative outcomes. Furthermore, we have applied SHapley Additive exPlanations (SHAP) to make the AI models interpretable. Finally, we analyzed the surgeries not converted to open versus those converted to open.

RESULTS

Overall, 585 patients and 22 variables were included. Multi-layer Perceptron (MLP) showed the highest performance for predicting surgery complexity and Random Forest (RF) for predicting postoperative outcomes. SHAP detected that MLP and RF gave the highest relevance to the variables "resection type" and "largest tumor size" for predicting surgery complexity and postoperative outcomes. In addition, we explored between surgeries converted to open and non-converted, finding statistically significant differences in the variables "tumor location," "blood loss," "complications," and "operation time."

CONCLUSION

We have observed how the application of SHAP allows us to understand the predictions of AI models in surgical complexity and the postoperative outcomes of laparoscopic liver surgery in segments 7 and 8.

Artificial intelligence in surgery

Artificial intelligence (AI) is rapidly emerging in healthcare, yet applications in surgery remain relatively nascent. Here we review the integration of AI in the field of surgery, centering our discussion on multifaceted improvements in surgical care in the preoperative, intraoperative and postoperative space. The emergence of foundation model architectures, wearable technologies and improving surgical data infrastructures is enabling rapid advances in AI interventions and utility. We discuss how maturing AI methods hold the potential to improve patient outcomes, facilitate surgical education and optimize surgical care. We review the current applications of deep learning approaches and outline a vision for future advances through multimodal foundation models.

Development and validation of 'Patient Optimizer' (POP) algorithms for predicting surgical risk with machine learning

BACKGROUND

Pre-operative risk assessment can help clinicians prepare patients for surgery, reducing the risk of perioperative complications, length of hospital stay, readmission and mortality. Further, it can facilitate collaborative decision-making and operational planning.

OBJECTIVE

To develop effective pre-operative risk assessment algorithms (referred to as Patient Optimizer or POP) using Machine Learning (ML) that predict the development of post-operative complications and provide pilot data to inform the design of a larger prospective study.

METHODS

After institutional ethics approval, we developed a base model that encapsulates the standard manual approach of combining patient-risk and procedure-risk. In an automated process, additional variables were included and tested with 10-fold cross-validation, and the best performing features were selected. The models were evaluated and confidence intervals calculated using bootstrapping. Clinical expertise was used to restrict the cardinality of categorical variables (e.g. pathology results) by including the most clinically relevant values. The models were created with logistic regression (LR) and extreme gradient-boosted trees using XGBoost (Chen and Guestrin, 2016). We evaluated performance using the area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC). Data was obtained from a metropolitan university teaching hospital from January 2015 to July 2020. Data collection was restricted to adult patients undergoing elective surgery.

RESULTS

A total of 11,475 adult admissions were included. The performance of XGBoost and LR was very similar across endpoints and metrics. For predicting the risk of any post-operative complication, kidney failure and length-of-stay (LOS), POP with XGBoost achieved an AUROC (95%CI) of 0.755 (0.744, 0.767), 0.869 (0.846, 0.891) and 0.841 (0.833, 0.847) respectively and AUPRC of 0.651 (0.632, 0.669), 0.336 (0.282, 0.390) and 0.741 (0.729, 0.753) respectively. For 30-day readmission and in-patient mortality, POP with XGBoost achieved an AUROC (95%CI) of 0.610 (0.587, 0.635) and 0.866 (0.777, 0.943) respectively and AUPRC of 0.116 (0.104, 0.132) and 0.031 (0.015, 0.072) respectively.

CONCLUSION

The POP algorithms effectively predicted any post-operative complication, kidney failure and LOS in the sample population. A larger study is justified to improve the algorithm to better predict complications and length of hospital stay. A larger dataset may also improve the prediction of additional specific complications, readmission and mortality.

Association between postinduction hypotension and postoperative mortality: a single-centre retrospective cohort study

PURPOSE

We aimed to elucidate whether postinduction hypotension (PIH), defined as hypotension between anesthesia induction and skin incision, and intraoperative hypotension (IOH) are associated with postoperative mortality.

METHODS

We conducted a retrospective cohort study of adult patients with an ASA Physical Status I-IV who underwent noncardiac and nonobstetric surgery under general anesthesia between 2015 and 2021 at Nagoya City University Hospital. The primary and secondary outcomes were 30-day and 90-day postoperative mortality, respectively. We calculated four hypotensive indices (with time proportion of the area under the threshold being the primary exposure variable) to evaluate the association between hypotension (defined as a mean blood pressure < 65 mm Hg) and mortality using multivariable logistic regression models. We used propensity score matching and RUSBoost (random under-sampling and boosting), a machine-learning model for imbalanced data, for sensitivity analyses.

RESULTS

Postinduction hypotension and IOH were observed in 82% and 84% of patients, respectively. The 30-day and 90-day postoperative mortality rates were 0.4% (52/14,210) and 1.0% (138/13,334), respectively. Postinduction hypotension was not associated with 30-day mortality (adjusted odds ratio [aOR], 1.03; 95% confidence interval [CI], 0.93 to 1.13; P = 0.60) and 90-day mortality (aOR, 1.01; 95% CI, 0.94 to 1.07; P = 0.82). Conversely, IOH was associated with 30-day mortality (aOR, 1.19; 95% CI, 1.12 to 1.27; P < 0.001) and 90-day mortality (aOR, 1.12; 95% CI, 1.06 to 1.19; P < 0.001). Sensitivity analyses supported the association of IOH but not PIH with postoperative mortality.

CONCLUSION

Despite limitations, including power and residual confounding, postoperative mortality was associated with IOH but not with PIH.

Development and validation of an interpretable Markov-embedded multilabel model for predicting risks of multiple postoperative complications among surgical inpatients: a multicenter prospective cohort study

BACKGROUND

When they encounter various highly related postoperative complications, existing risk evaluation tools that focus on single or any complications are inadequate in clinical practice. This seriously hinders complication management because of the lack of a quantitative basis. An interpretable multilabel model framework that predicts multiple complications simultaneously is urgently needed.

MATERIALS AND METHODS

The authors included 50 325 inpatients from a large multicenter cohort (2014-2017). The authors separated patients from one hospital for external validation and randomly split the remaining patients into training and internal validation sets. A MARKov-EmbeDded (MARKED) multilabel model was proposed, and three models were trained for comparison: binary relevance, a fully connected network (FULLNET), and a deep neural network. Performance was mainly evaluated using the area under the receiver operating characteristic curve (AUC). The authors interpreted the model using Shapley Additive Explanations. Complication-specific risk and risk source inference were provided at the individual level.

RESULTS

There were 26 292, 6574, and 17 459 inpatients in the training, internal validation, and external validation sets, respectively. For the external validation set, MARKED achieved the highest average AUC (0.818, 95% CI: 0.771-0.864) across eight outcomes [compared with binary relevance, 0.799 (0.748-0.849), FULLNET, 0.806 (0.756-0.856), and deep neural network, 0.815 (0.765-0.866)]. Specifically, the AUCs of MARKED were above 0.9 for cardiac complications [0.927 (0.894-0.960)], neurological complications [0.905 (0.870-0.941)], and mortality [0.902 (0.867-0.937)]. Serum albumin, surgical specialties, emergency case, American Society of Anesthesiologists score, age, and sex were the six most important preoperative variables. The interaction between complications contributed more than the preoperative variables, and formed a hierarchical chain of risk factors, mild complications, and severe complications.

CONCLUSION

The authors demonstrated the advantage of MARKED in terms of performance and interpretability. The authors expect that the identification of high-risk patients and the inference of the risk source for specific complications will be valuable for clinical decision-making.

Prediction of Complications and Prognostication in Perioperative Medicine: A Systematic Review and PROBAST Assessment of Machine Learning Tools

BACKGROUND

The utilization of artificial intelligence and machine learning as diagnostic and predictive tools in perioperative medicine holds great promise. Indeed, many studies have been performed in recent years to explore the potential. The purpose of this systematic review is to assess the current state of machine learning in perioperative medicine, its utility in prediction of complications and prognostication, and limitations related to bias and validation.

METHODS

A multidisciplinary team of clinicians and engineers conducted a systematic review using the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) protocol. Multiple databases were searched, including Scopus, Cumulative Index to Nursing and Allied Health Literature (CINAHL), the Cochrane Library, PubMed, Medline, Embase, and Web of Science. The systematic review focused on study design, type of machine learning model used, validation techniques applied, and reported model performance on prediction of complications and prognostication. This review further classified outcomes and machine learning applications using an ad hoc classification system. The Prediction model Risk Of Bias Assessment Tool (PROBAST) was used to assess risk of bias and applicability of the studies.

RESULTS

A total of 103 studies were identified. The models reported in the literature were primarily based on single-center validations (75%), with only 13% being externally validated across multiple centers. Most of the mortality models demonstrated a limited ability to discriminate and classify effectively. The PROBAST assessment indicated a high risk of systematic errors in predicted outcomes and artificial intelligence or machine learning applications.

CONCLUSIONS

The findings indicate that the development of this field is still in its early stages. This systematic review indicates that application of machine learning in perioperative medicine is still at an early stage. While many studies suggest potential utility, several key challenges must be first overcome before their introduction into clinical practice.

EDITOR’S PERSPECTIVE

Development and validation of artificial intelligence models to predict urinary tract infections and secondary bloodstream infections in adult patients

BACKGROUND

Traditional culture methods are time-consuming, making it difficult to utilize the results in the early stage of urinary tract infection (UTI) management, and automated urinalyses alone show insufficient performance for diagnosing UTIs. Several models have been proposed to predict urine culture positivity based on urinalysis. However, most of them have not been externally validated or consisted solely of urinalysis data obtained using one specific commercial analyzer.

METHODS

A total of 259,187 patients were enrolled to develop artificial intelligence (AI) models. AI models were developed and validated for the diagnosis of UTI and urinary tract related-bloodstream infection (UT-BSI). The predictive performance of conventional urinalysis and AI algorithms were assessed by the areas under the receiver operating characteristic curve (AUROC). We also visualized feature importance rankings as Shapley additive explanation bar plots.

RESULTS

In the two cohorts, the positive rates of urine culture tests were 25.2% and 30.4%, and the proportions of cases classified as UT-BSI were 1.8% and 1.6%. As a result of predicting UTI from the automated urinalysis, the AUROC were 0.745 (0.743-0.746) and 0.740 (0.737-0.743), and most AI algorithms presented excellent discriminant performance (AUROC > 0.9). In the external validation dataset, the XGBoost model achieved the best values in predicting both UTI (AUROC 0.967 [0.966-0.968]) and UT-BSI (AUROC 0.955 [0.951-0.959]). A reduced model using ten parameters was also derived.

CONCLUSIONS

We found that AI models can improve the early prediction of urine culture positivity and UT-BSI by combining automated urinalysis with other clinical information. Clinical utilization of the model can reduce the risk of delayed antimicrobial therapy in patients with nonspecific symptoms of UTI and classify patients with UT-BSI who require further treatment and close monitoring.

Integration of machine learning to identify diagnostic genes in leukocytes for acute myocardial infarction patients

BACKGROUND

Acute myocardial infarction (AMI) has two clinical characteristics: high missed diagnosis and dysfunction of leukocytes. Transcriptional RNA on leukocytes is closely related to the course evolution of AMI patients. We hypothesized that transcriptional RNA in leukocytes might provide potential diagnostic value for AMI. Integration machine learning (IML) was first used to explore AMI discrimination genes. The following clinical study was performed to validate the results.

METHODS

A total of four AMI microarrays (derived from the Gene Expression Omnibus) were included in bioanalysis (220 sample size). Then, the clinical validation was finished with 20 AMI and 20 stable coronary artery disease patients (SCAD). At a ratio of 5:2, GSE59867 was included in the training set, while GSE60993, GSE62646, and GSE48060 were included in the testing set. IML was explicitly proposed in this research, which is composed of six machine learning algorithms, including support vector machine (SVM), neural network (NN), random forest (RF), gradient boosting machine (GBM), decision trees (DT), and least absolute shrinkage and selection operator (LASSO). IML had two functions in this research: filtered optimized variables and predicted the categorized value. Finally, The RNA of the recruited patients was analyzed to verify the results of IML.

RESULTS

Thirty-nine differentially expressed genes (DEGs) were identified between controls and AMI individuals from the training sets. Among the thirty-nine DEGs, IML was used to process the predicted classification model and identify potential candidate genes with overall normalized weights > 1. Finally, two genes (AQP9 and SOCS3) show their diagnosis value with the area under the curve (AUC) > 0.9 in both the training and testing sets. The clinical study verified the significance of AQP9 and SOCS3. Notably, more stenotic coronary arteries or severe Killip classification indicated higher levels of these two genes, especially SOCS3. These two genes correlated with two immune cell types, monocytes and neutrophils.

CONCLUSION

AQP9 and SOCS3 in leukocytes may be conducive to identifying AMI patients with SCAD patients. AQP9 and SOCS3 are closely associated with monocytes and neutrophils, which might contribute to advancing AMI diagnosis and shed light on novel genetic markers. Multiple clinical characteristics, multicenter, and large-sample relevant trials are still needed to confirm its clinical value.

Preoperative shock index in major abdominal emergency surgery

Introduction

Major abdominal emergency surgery (MAES) patients have a high risk of mortality and complications. The time-sensitive nature of MAES necessitates an easily calculable risk-scoring tool. Shock index (SI) is obtained by dividing heart rate (HR) by systolic blood pressure (SBP) and provides insight into a patient's haemodynamic status. We aimed to evaluate SI's usefulness in predicting postoperative mortality, acute kidney injury (AKI), requirements for intensive care unit (ICU) and high-dependency monitoring, and the ICU length of stay (LOS).

Method

We retrospectively reviewed 212,089 MAES patients from January 2013 to December 2020. The cohort was propensity matched, and 3960 patients were included. The first HR and SBP recorded in the anaesthesia chart were used to calculate SI. Regression models were used to investigate the association between SI and outcomes. The relationship between SI and survival was explored with Kaplan-Meier curves.

Results

There were significant associations between SI and mortality at 1 month (odds ratio [OR] 2.40 [1.67-3.39], P<0.001), 3 months (OR 2.13 [1.56-2.88], P<0.001), and at 2 years (OR 1.77 [1.38-2.25], P<0.001). Multivariate analysis revealed significant relationships between SI and mortality at 1 month (OR 3.51 [1.20-10.3], P=0.021) and at 3 months (OR 3.05 [1.07-8.54], P=0.034). Univariate and multivariate analysis also revealed significant relationships between SI and AKI (P<0.001), postoperative ICU admission (P<0.005) and ICU LOS (P<0.001). SI does not significantly affect 2-year mortality.

Conclusion

SI is useful in predicting postopera-tive mortality at 1 month, 3 months, AKI, postoperative ICU admission and ICU LOS.

A Comprehensive Analysis and Review of Artificial Intelligence in Anaesthesia

In the field of anaesthesia, artificial intelligence (AI) has become a game-changing technology. Applications of AI include keeping records, monitoring patients, calculating and administering drugs, and carrying out mechanical procedures. This article explores the current uses, challenges, and prospective applications of AI in anaesthesia practices. This review discusses AI-supported systems like anaesthesia information management systems (AIMS), mechanical robots for carrying out procedures, and pharmacological models for drug delivery. AIMS has helped in automated record-keeping, predicting bad events, and monitoring the vital signs of the patient. Their application has a vital role in improving the efficacy of anaesthesia management and patient safety. The application of AI in anaesthesia comes with its own unique difficulties. Noteworthy obstacles include issues with data quantity and quality, technical limitations, and moral and legal dilemmas. The key to overcoming these barriers is to set guidelines for the ethical use of AI in healthcare, improve the reliability and comprehension of AI systems, and certify the health data precision and security. AI has very bright potential. Exciting future directions include developments in AI and machine learning thus development of new applications, and the possible enhancement in training and education. Potential research areas include the application of AI to chronic disease management, pain management, and the reinforcement of anaesthesiologists' education. AI could be used to design authentic lifelike training simulations and individualized student feedback systems, hence transforming anaesthesia education and training methodology. For this review, we conducted a PubMed, Google Scholar, and Cochrane Database search in 2022-2023 and retrieved articles on AI and its uses in anaesthesia. Recommendations for future research and development include strengthening the safety and reliability of health data, building a better understanding of AI systems, and looking into new areas of use. The power of AI can be used to innovate anaesthesia practices by concentrating on these areas.

Is a guideline required to predict the intensive care unit need of patients over 65 years of age during the pre-operative period? A comparison of the American Society of Anesthesiologists, lung ultrasound score, Charlson age-added comorbidity index, surgi

BACKGROUND

All pre-operative, intra-operative, and post-operative variables of the patients at 65 years of age who had a surgi-cal procedure determine the necessity of post-operative intensive care unit (ICU) monitoring. The indication for post-operative ICU is detected through ideal scoring systems related to the surgery and anesthesia that will be performed easily and fast would prevent the development of morbidity and mortality in high-risk patients. In the present study, we compared the efficacy of the American Society of Anesthesiologists (ASA) score, lung ultrasound score (LUSS), Charlson age-added comorbidity index (CACI), and surgical outcome risk tool (SORT) score of the indication for ICU. The hypothesis of our study is to show that real visual LUSS is superior to the screening test SORT, CACI, and the other score, ASA, for ICU indication determination.

METHODS

The study enrolled 101 patients over 65 years of age who will have surgical procedures under elective conditions. De-mographic features, clinical parameters, ICU indications, ASA, LUSS, CACI, and SORTs of the patients were calculated prospectively and recorded. The effects of patients' ASA, LUSS, CACI, and SORT on determining the need for postoperative ICU admission were examined.

RESULTS

The age of patients who needed post-operative ICU admission was significantly higher than those who did not need post-operative ICU admission (P<0.001). The groups did not show differences in terms of gender, body mass index, smoking, and type of anesthesia (P>0.05). ASA, LUSS, CACI, and SORT were significantly higher for patients who needed post-operative ICU admission (P<0.001). The proportion of patients who needed post-operative ICU admission was higher for patients with post-operative ICU indication (P<0.001). The number of consultations was significantly higher for patients who needed post-operative ICU admission (P<0.001). SORT was found to be the highest accuracy for predicting the need for post-operative ICU admission.

CONCLUSION

It was detected that ASA, LUSS, CACI, and SORT are effective for the determination of the ICU indication in the pre-operative evaluation process of patients over the age of 65 who had elective surgery. However, the efficiency of SORT was found to be superior to the others.

Artificial Intelligence Applications in Space Medicine

INTRODUCTION:During future interplanetary space missions, a number of health conditions may arise, owing to the hostile environment of space and the myriad of stressors experienced by the crew. When managing these conditions, crews will be required to make accurate, timely clinical decisions at a high level of autonomy, as telecommunication delays and increasing distances restrict real-time support from the ground. On Earth, artificial intelligence (AI) has proven successful in healthcare, augmenting expert clinical decision-making or enhancing medical knowledge where it is lacking. Similarly, deploying AI tools in the context of a space mission could improve crew self-reliance and healthcare delivery.METHODS: We conducted a narrative review to discuss existing AI applications that could improve the prevention, recognition, evaluation, and management of the most mission-critical conditions, including psychological and mental health, acute radiation sickness, surgical emergencies, spaceflight-associated neuro-ocular syndrome, infections, and cardiovascular deconditioning.RESULTS: Some examples of the applications we identified include AI chatbots designed to prevent and mitigate psychological and mental health conditions, automated medical imaging analysis, and closed-loop systems for hemodynamic optimization. We also discuss at length gaps in current technologies, as well as the key challenges and limitations of developing and deploying AI for space medicine to inform future research and innovation. Indeed, shifts in patient cohorts, space-induced physiological changes, limited size and breadth of space biomedical datasets, and changes in disease characteristics may render the models invalid when transferred from ground settings into space.Cheung HC, De Louche C, Komorowski M. Artificial intelligence applications in space medicine. Aerosp Med Hum Perform. 2023; 94(8):610-622.

Artificial Intelligence-enabled Decision Support in Surgery: State-of-the-art and Future Directions

OBJECTIVE

To summarize state-of-the-art artificial intelligence-enabled decision support in surgery and to quantify deficiencies in scientific rigor and reporting.

BACKGROUND

To positively affect surgical care, decision-support models must exceed current reporting guideline requirements by performing external and real-time validation, enrolling adequate sample sizes, reporting model precision, assessing performance across vulnerable populations, and achieving clinical implementation; the degree to which published models meet these criteria is unknown.

METHODS

Embase, PubMed, and MEDLINE databases were searched from their inception to September 21, 2022 for articles describing artificial intelligence-enabled decision support in surgery that uses preoperative or intraoperative data elements to predict complications within 90 days of surgery. Scientific rigor and reporting criteria were assessed and reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews guidelines.

RESULTS

Sample size ranged from 163-2,882,526, with 8/36 articles (22.2%) featuring sample sizes of less than 2000; 7 of these 8 articles (87.5%) had below-average (<0.83) area under the receiver operating characteristic or accuracy. Overall, 29 articles (80.6%) performed internal validation only, 5 (13.8%) performed external validation, and 2 (5.6%) performed real-time validation. Twenty-three articles (63.9%) reported precision. No articles reported performance across sociodemographic categories. Thirteen articles (36.1%) presented a framework that could be used for clinical implementation; none assessed clinical implementation efficacy.

CONCLUSIONS

Artificial intelligence-enabled decision support in surgery is limited by reliance on internal validation, small sample sizes that risk overfitting and sacrifice predictive performance, and failure to report confidence intervals, precision, equity analyses, and clinical implementation. Researchers should strive to improve scientific quality.

Exploring Artificial Intelligence in Anesthesia: A Primer on Ethics, and Clinical Applications

The field of anesthesia has always been at the forefront of innovation and technology, and the integration of Artificial Intelligence (AI) represents the next frontier in anesthesia care. The use of AI and its subtypes, such as machine learning, has the potential to improve efficiency, reduce costs, and ameliorate patient outcomes. AI can assist with decision making, but its primary advantage lies in empowering anesthesiologists to adopt a proactive approach to address clinical issues. The potential uses of AI in anesthesia can be schematically grouped into clinical decision support and pharmacologic and mechanical robotic applications. Tele-anesthesia includes strategies of telemedicine, as well as device networking, for improving logistics in the operating room, and augmented reality approaches for training and assistance. Despite the growing scientific interest, further research and validation are needed to fully understand the benefits and limitations of these applications in clinical practice. Moreover, the ethical implications of AI in anesthesia must also be considered to ensure that patient safety and privacy are not compromised. This paper aims to provide a comprehensive overview of AI in anesthesia, including its current and potential applications, and the ethical considerations that must be considered to ensure the safe and effective use of the technology.

Transforming the perioperative medicine care model: The Singapore experience

More than 300 million surgeries are performed worldwide annually. Established perioperative centres in the UK, USA and Australia have demonstrated the impact of improving perioperative care in reducing costs, increasing patient satisfaction and improving population health. Likewise, the surgical burden of care in Asia is increasing, but with sociocultural, economic and epigenetic differences compared to the west. As Singapore's largest hospital, the Singapore General Hospital pre-admission perioperative clinic sees about 20,000 patients annually. We aim to illustrate Singapore General Hospital's perioperative model of care to contribute to the paucity of literature describing perioperative programme implementation within Asia, and to encourage the cross-sharing of perioperative practices internationally. Our perioperative framework navigates risk assessment, risk counselling, and mitigation of health, medical and functional risks to better patients' perioperative outcomes and population health. We have implemented evidence-based pathways for common conditions such as anaemia and malnutrition, including a multidisciplinary programme for the elderly to tackle frailty and reduce length of stay. We describe how we have enhanced local risk profiling with the Combined Assessment of Risk Encountered in Surgery surgical risk calculator derived locally using a gradient boosting machine learning model. Finally, we report clinical outcomes of these interventions and discuss further challenges and new initiatives at each tier of our perioperative model. Our perioperative care model provides a framework that other centres can adopt to promote value-driven care, while catering for differences in the Asian population, thereby promoting evidence-based improvements in the area of perioperative medicine.

Dynamic predictions of postoperative complications from explainable, uncertainty-aware, and multi-task deep neural networks

Accurate prediction of postoperative complications can inform shared decisions regarding prognosis, preoperative risk-reduction, and postoperative resource use. We hypothesized that multi-task deep learning models would outperform conventional machine learning models in predicting postoperative complications, and that integrating high-resolution intraoperative physiological time series would result in more granular and personalized health representations that would improve prognostication compared to preoperative predictions. In a longitudinal cohort study of 56,242 patients undergoing 67,481 inpatient surgical procedures at a university medical center, we compared deep learning models with random forests and XGBoost for predicting nine common postoperative complications using preoperative, intraoperative, and perioperative patient data. Our study indicated several significant results across experimental settings that suggest the utility of deep learning for capturing more precise representations of patient health for augmented surgical decision support. Multi-task learning improved efficiency by reducing computational resources without compromising predictive performance. Integrated gradients interpretability mechanisms identified potentially modifiable risk factors for each complication. Monte Carlo dropout methods provided a quantitative measure of prediction uncertainty that has the potential to enhance clinical trust. Multi-task learning, interpretability mechanisms, and uncertainty metrics demonstrated potential to facilitate effective clinical implementation.

Multiple-model machine learning identifies potential functional genes in dilated cardiomyopathy

Introduction

Machine learning (ML) has gained intensive popularity in various fields, such as disease diagnosis in healthcare. However, it has limitation for single algorithm to explore the diagnosing value of dilated cardiomyopathy (DCM). We aim to develop a novel overall normalized sum weight of multiple-model MLs to assess the diagnosing value in DCM.

Methods

Gene expression data were selected from previously published databases (six sets of eligible microarrays, 386 samples) with eligible criteria. Two sets of microarrays were used as training; the others were studied in the testing sets (ratio 5:1). Totally, we identified 20 differently expressed genes (DEGs) between DCM and control individuals (7 upregulated and 13 down-regulated).

Results

We developed six classification ML methods to identify potential candidate genes based on their overall weights. Three genes, serine proteinase inhibitor A3 (SERPINA3), frizzled-related proteins (FRPs) 3 (FRZB), and ficolin 3 (FCN3) were finally identified as the receiver operating characteristic (ROC). Interestingly, we found all three genes correlated considerably with plasma cells. Importantly, not only in training sets but also testing sets, the areas under the curve (AUCs) for SERPINA3, FRZB, and FCN3 were greater than 0.88. The ROC of SERPINA3 was significantly high (0.940 in training and 0.918 in testing sets), indicating it is a potentially functional gene in DCM. Especially, the plasma levels in DCM patients of SERPINA3, FCN, and FRZB were significant compared with healthy control.

Discussion

SERPINA3, FRZB, and FCN3 might be potential diagnosis targets for DCM, Further verification work could be implemented.

Present and future of machine learning in breast surgery: systematic review

BACKGROUND

Machine learning is a set of models and methods that can automatically detect patterns in vast amounts of data, extract information, and use it to perform decision-making under uncertain conditions. The potential of machine learning is significant, and breast surgeons must strive to be informed with up-to-date knowledge and its applications.

METHODS

A systematic database search of Embase, MEDLINE, the Cochrane database, and Google Scholar, from inception to December 2021, was conducted of original articles that explored the use of machine learning and/or artificial intelligence in breast surgery in EMBASE, MEDLINE, Cochrane database and Google Scholar.

RESULTS

The search yielded 477 articles, of which 14 studies were included in this review, featuring 73 847 patients. Four main areas of machine learning application were identified: predictive modelling of surgical outcomes; breast imaging-based context; screening and triaging of patients with breast cancer; and as network utility for detection. There is evident value of machine learning in preoperative planning and in providing information for surgery both in a cancer and an aesthetic context. Machine learning outperformed traditional statistical modelling in all studies for predicting mortality, morbidity, and quality of life outcomes. Machine learning patterns and associations could support planning, anatomical visualization, and surgical navigation.

CONCLUSION

Machine learning demonstrated promising applications for improving breast surgery outcomes and patient-centred care. Neveretheless, there remain important limitations and ethical concerns relating to implementing artificial intelligence into everyday surgical practices.

Artificial intelligence-based preoperative prediction system for diagnosis and prognosis in epithelial ovarian cancer: A multicenter study

Background

Ovarian cancer (OC) is the most lethal gynecological malignancy, with limited early screening methods and poor prognosis. Artificial intelligence technology has made a great breakthrough in cancer diagnosis.

Purpose

We aim to develop a specific interpretable machine learning (ML) prediction model for the diagnosis and prognosis of epithelial ovarian cancer (EOC) based on a variety of biomarkers.

Methods

A total of 521 patients with EOC and 144 patients with benign gynecological diseases were enrolled including derivation datasets and an external validation cohort. The predicted information was acquired by 9 supervised ML methods, through 34 parameters. Behind predicted reasons for the best ML were improved by using the SHapley Additive exPlanations (SHAP) algorithm. In addition, the prognosis of EOC was analyzed by unsupervised clustering and Kaplan–Meier (KM) survival analysis.

Results

ML technology was superior to conventional logistic regression in predicting EOC diagnosis and XGBoost performed best in the external validation datasets. The AUC values of distinguishing EOC and benign disease patients, determining pathological type, grade and clinical stage were 0.958 (0.926-0.989), 0.792 (0.701-0.8834), 0.819 (0.687-0.950) and 0.68 (0.573-0.788) respectively. For negative CA-125 EOC patients, the AUC performance of XGBoost model was 0.835(0.763-0.907). We used unsupervised cluster analysis to identify EOC subgroups with significantly poor overall survival (p-value <0.0001) and recurrence-free survival (p-value <0.0001).

Conclusions

Based on the preoperative characteristics, we proved that ML algorithm can provide an acceptable diagnosis and prognosis prediction model for EOC patients. Meanwhile, SHAP analysis can improve the interpretability of ML models and contribute to precision medicine.

Machine learning prediction of postoperative major adverse cardiovascular events in geriatric patients: a prospective cohort study

Background

Postoperative major adverse cardiovascular events (MACEs) account for more than one-third of perioperative deaths. Geriatric patients are more vulnerable to postoperative MACEs than younger patients. Identifying high-risk patients in advance can help with clinical decision making and improve prognosis. This study aimed to develop a machine learning model for the preoperative prediction of postoperative MACEs in geriatric patients.

Methods

We collected patients’ clinical data and laboratory tests prospectively. All patients over 65 years who underwent surgeries in West China Hospital of Sichuan University from June 25, 2019 to June 29, 2020 were included. Models based on extreme gradient boosting (XGB), gradient boosting machine, random forest, support vector machine, and Elastic Net logistic regression were trained. The models’ performance was compared according to area under the precision-recall curve (AUPRC), area under the receiver operating characteristic curve (AUROC) and Brier score. To minimize the influence of clinical intervention, we trained the model based on undersampling set. Variables with little contribution were excluded to simplify the model for ensuring the ease of use in clinical settings.

Results

We enrolled 5705 geriatric patients into the final dataset. Of those patients, 171 (3.0%) developed postoperative MACEs within 30 days after surgery. The XGB model outperformed other machine learning models with AUPRC of 0.404(95% confidence interval [CI]: 0.219–0.589), AUROC of 0.870(95%CI: 0.786–0.938) and Brier score of 0.024(95% CI: 0.016–0.032). Model trained on undersampling set showed improved performance with AUPRC of 0.511(95% CI: 0.344–0.667, p < 0.001), AUROC of 0.912(95% CI: 0.847–0.962, p < 0.001) and Brier score of 0.020 (95% CI: 0.013–0.028, p < 0.001). After removing variables with little contribution, the undersampling model showed comparable predictive accuracy with AUPRC of 0.507(95% CI: 0.338–0.669, p = 0.36), AUROC of 0.896(95%CI: 0.826–0.953, p < 0.001) and Brier score of 0.020(95% CI: 0.013–0.028, p = 0.20).

Conclusions

In this prospective study, we developed machine learning models for preoperative prediction of postoperative MACEs in geriatric patients. The XGB model showed the best performance. Undersampling method achieved further improvement of model performance.

Trial registration

The protocol of this study was registered at www.chictr.org.cn (15/08/2019, ChiCTR1900025160)

Supplementary Information

The online version contains supplementary material available at 10.1186/s12871-022-01827-x.

A multicenter prospective study on postoperative pulmonary complications prediction in geriatric patients with deep neural network model

Aim

Postoperative pulmonary complications (PPCs) can increase the risk of postoperative mortality, and the geriatric population has high incidence of PPCs. Early identification of high-risk geriatric patients is of great value for clinical decision making and prognosis improvement. Existing prediction models are based purely on structured data, and they lack predictive accuracy in geriatric patients. We aimed to develop and validate a deep neural network model based on combined natural language data and structured data for improving the prediction of PPCs in geriatric patients.

Methods

We consecutively enrolled patients aged ≥65 years who underwent surgery under general anesthesia at seven hospitals in China. Data from the West China Hospital of Sichuan University were used as the derivation dataset, and a deep neural network model was developed based on combined natural language data and structured data. Data from the six other hospitals were combined for external validation.

Results

The derivation dataset included 12,240 geriatric patients, and 1949(15.9%) patients developed PPCs. Our deep neural network model outperformed other machine learning models with an area under the precision-recall curve (AUPRC) of 0.657(95% confidence interval [CI], 0.655–0.658) and an area under the receiver operating characteristic curve (AUROC) of 0.884(95% CI, 0.883–0.885). The external dataset included 7579 patients, and 776(10.2%) patients developed PPCs. In external validation, the AUPRC was 0.632(95%CI, 0.632–0.633) and the AUROC was 0.889(95%CI, 0.888–0.889).

Conclusions

This study indicated that the deep neural network model based on combined natural language data and structured data could improve the prediction of PPCs in geriatric patients.

Machine learning-assisted prediction of pneumonia based on non-invasive measures

Background

Pneumonia is an infection of the lungs that is characterized by high morbidity and mortality. The use of machine learning systems to detect respiratory diseases via non-invasive measures such as physical and laboratory parameters is gaining momentum and has been proposed to decrease diagnostic uncertainty associated with bacterial pneumonia. Herein, this study conducted several experiments using eight machine learning models to predict pneumonia based on biomarkers, laboratory parameters, and physical features.

Methods

We perform machine-learning analysis on 535 different patients, each with 45 features. Data normalization to rescale all real-valued features was performed. Since it is a binary problem, we categorized each patient into one class at a time. We designed three experiments to evaluate the models: (1) feature selection techniques to select appropriate features for the models, (2) experiments on the imbalanced original dataset, and (3) experiments on the SMOTE data. We then compared eight machine learning models to evaluate their effectiveness in predicting pneumonia

Results

Biomarkers such as C-reactive protein and procalcitonin demonstrated the most significant discriminating power. Ensemble machine learning models such as RF (accuracy = 92.0%, precision = 91.3%, recall = 96.0%, f1-Score = 93.6%) and XGBoost (accuracy = 90.8%, precision = 92.6%, recall = 92.3%, f1-score = 92.4%) achieved the highest performance accuracy on the original dataset with AUCs of 0.96 and 0.97, respectively. On the SMOTE dataset, RF and XGBoost achieved the highest prediction results with f1-scores of 92.0 and 91.2%, respectively. Also, AUC of 0.97 was achieved for both RF and XGBoost models.

Conclusions

Our models showed that in the diagnosis of pneumonia, individual clinical history, laboratory indicators, and symptoms do not have adequate discriminatory power. We can also conclude that the ensemble ML models performed better in this study.

Preoperative prediction of the need for arterial and central venous catheterization using machine learning techniques

Some surgical patients require an arterial or central venous catheterization intraoperatively. This decision relied solely on the experience of individual anesthesiologists; however, these decisions are not easy for clinicians who are in an emergency or inexperienced. Therefore, applying recent artificial intelligence techniques to automatically extractable data from electronic medical record (EMR) could create a very clinically useful model in this situation. This study aimed to develop a model that is easy to apply in real clinical settings by implementing a prediction model for the preoperative decision to insert an arterial and central venous catheter and that can be automatically linked to the EMR. We collected and retrospectively analyzed data from 66,522 patients, > 18 years of age, who underwent non-cardiac surgeries from March 2019 to April 2021 at the single tertiary medical center. Data included demographics, pre-operative laboratory tests, surgical information, and catheterization information. When compared with other machine learning methods, the DNN model showed the best predictive performance in terms of the area under receiver operating characteristic curve and area under the precision-recall curve. Operation code information accounted for the largest portion of the prediction. This can be applied to clinical fields using operation code and minimal preoperative clinical information.

Multi-center validation of machine learning model for preoperative prediction of postoperative mortality

Accurate prediction of postoperative mortality is important for not only successful postoperative patient care but also for information-based shared decision-making with patients and efficient allocation of medical resources. This study aimed to create a machine-learning prediction model for 30-day mortality after a non-cardiac surgery that adapts to the manageable amount of clinical information as input features and is validated against multi-centered rather than single-centered data. Data were collected from 454,404 patients over 18 years of age who underwent non-cardiac surgeries from four independent institutions. We performed a retrospective analysis of the retrieved data. Only 12–18 clinical variables were used for model training. Logistic regression, random forest classifier, extreme gradient boosting (XGBoost), and deep neural network methods were applied to compare the prediction performances. To reduce overfitting and create a robust model, bootstrapping and grid search with tenfold cross-validation were performed. The XGBoost method in Seoul National University Hospital (SNUH) data delivers the best performance in terms of the area under receiver operating characteristic curve (AUROC) (0.9376) and the area under the precision-recall curve (0.1593). The predictive performance was the best when the SNUH model was validated with Ewha Womans University Medical Center data (AUROC, 0.941). Preoperative albumin, prothrombin time, and age were the most important features in the model for each hospital. It is possible to create a robust artificial intelligence prediction model applicable to multiple institutions through a light predictive model using only minimal preoperative information that can be automatically extracted from each hospital.

POTTER-ICU: An artificial intelligence smartphone-accessible tool to predict the need for intensive care after emergency surgery

BACKGROUND

Delays in admitting high-risk emergency surgery patients to the intensive care unit result in worse outcomes and increased health care costs. We aimed to use interpretable artificial intelligence technology to create a preoperative predictor for postoperative intensive care unit need in emergency surgery patients.

METHODS

A novel, interpretable artificial intelligence technology called optimal classification trees was leveraged in an 80:20 train:test split of adult emergency surgery patients in the 2007-2017 American College of Surgeons National Surgical Quality Improvement Program database. Demographics, comorbidities, and laboratory values were used to develop, train, and then validate optimal classification tree algorithms to predict the need for postoperative intensive care unit admission. The latter was defined as postoperative death or the development of 1 or more postoperative complications warranting critical care (eg, unplanned intubation, ventilator requirement ≥48 hours, cardiac arrest requiring cardiopulmonary resuscitation, and septic shock). An interactive and user-friendly application was created. C statistics were used to measure performance.

RESULTS

A total of 464,861 patients were included. The mean age was 55 years, 48% were male, and 11% developed severe postoperative complications warranting critical care. The Predictive OpTimal Trees in Emergency Surgery Risk Intensive Care Unit application was created as the user-friendly interface of the complex optimal classification tree algorithms. The number of questions (ie, tree depths) needed to predict intensive care unit admission ranged from 2 to 11. The Predictive OpTimal Trees in Emergency Surgery Risk Intensive Care Unit application had excellent discrimination for predicting the need for intensive care unit admission (C statistics: 0.89 train, 0.88 test).

CONCLUSION

We recommend the Predictive OpTimal Trees in Emergency Surgery Risk Intensive Care Unit application as an accurate, artificial intelligence-based tool for predicting severe complications warranting intensive care unit admission after emergency surgery. The Predictive OpTimal Trees in Emergency Surgery Risk Intensive Care Unit application can prove useful to triage patients to the intensive care unit and to potentially decrease failure to rescue in emergency surgery patients.

Risk Factors of Severe Clostridioides difficile Infection; Sequential Organ Failure Assessment Score, Antibiotics, and Ribotypes

We aimed to determine whether the Sequential Organ Failure Assessment (SOFA) score predicts the prognosis of patients with Clostridioides difficile infection (CDI). In addition, the association between the type of antibiotic used and PCR ribotypes was analyzed. We conducted a propensity score (PS)-matched study and machine learning analysis using clinical data from all adult patients with confirmed CDI in three South Korean hospitals. A total of 5,337 adult patients with CDI were included in this study, and 828 (15.5%) were classified as having severe CDI. The top variables selected by the machine learning models were maximum body temperature, platelet count, eosinophil count, oxygen saturation, Glasgow Coma Scale, serum albumin, and respiratory rate. After propensity score-matching, the SOFA score, white blood cell (WBC) count, serum albumin level, and ventilator use were significantly associated with severe CDI (P < 0.001 for all). The log-rank test of SOFA score ≥ 4 significantly differentiated severe CDI patients from the non-severe group. The use of fluoroquinolone was more related to CDI patients with ribotype 018 strains than to ribotype 014/020 (P < 0.001). Even after controlling for other variables using propensity score matching analysis, we found that the SOFA score was a clinical predictor of severe CDI. We also demonstrated that the use of fluoroquinolones in hospital settings could be associated with the PCR ribotype in patients with CDI.

Mortality prediction of patients in intensive care units using machine learning algorithms based on electronic health records

Improving predictive models for intensive care unit (ICU) inpatients requires a new strategy that periodically includes the latest clinical data and can be updated to reflect local characteristics. We extracted data from all adult patients admitted to the ICUs of two university hospitals with different characteristics from 2006 to 2020, and a total of 85,146 patients were included in this study. Machine learning algorithms were trained to predict in-hospital mortality. The predictive performance of conventional scoring models and machine learning algorithms was assessed by the area under the receiver operating characteristic curve (AUROC). The conventional scoring models had various predictive powers, with the SAPS III (AUROC 0.773 [0.766-0.779] for hospital S) and APACHE III (AUROC 0.803 [0.795-0.810] for hospital G) showing the highest AUROC among them. The best performing machine learning models achieved an AUROC of 0.977 (0.973-0.980) in hospital S and 0.955 (0.950-0.961) in hospital G. The use of ML models in conjunction with conventional scoring systems can provide more useful information for predicting the prognosis of critically ill patients. In this study, we suggest that the predictive model can be made more robust by training with the individual data of each hospital.

Discordant American Society of Anesthesiologists Physical Status Classification between anesthesiologists and surgeons and its correlation with adverse patient outcomes

The American Society of Anesthesiologists Physical Status Classification (ASA) is used for communication of patient health status, risk scoring, benchmarking and financial claims. Prior studies using hypothetical scenarios have shown poor concordance of ASA classification among healthcare providers. There is a paucity of studies using clinical data, and of clinical factors or patient outcomes associated with discordant classification. The study aims to assess ASA classification concordance between surgeons and anesthesiologists, factors surrounding discordance and its impact on patient outcomes. This retrospective cohort study was conducted in a tertiary medical center on 46,284 consecutive patients undergoing elective surgery between January 2017 and December 2019. The ASA class showed moderate concordance (weighted Cohen’s κ 0.53) between surgeons and anesthesiologists. We found significant associations between discordant classification and patient comorbidities, age and race. Patients with discordant classification had a higher risk of 30-day mortality (odds ratio (OR) 2.00, 95% confidence interval (CI) = 1.52–2.62, p < 0.0001), 1-year mortality (OR 1.53, 95% CI = 1.38–1.69, p < 0.0001), and Intensive Care Unit admission > 24 h (OR 1.69, 95% CI = 1.47–1.94, p < 0.0001). Hence, there is a need for improved standardization of ASA scoring and cross-specialty review in ASA-discordant cases.

Development and Validation of a Multivariable Prediction Model for Postoperative Intensive Care Unit Stay in a Broad Surgical Population

IMPORTANCE

Despite limited capacity and expensive cost, there are minimal objective data to guide postoperative allocation of intensive care unit (ICU) beds. The Surgical Risk Preoperative Assessment System (SURPAS) uses 8 preoperative variables to predict many common postoperative complications, but it has not yet been evaluated in predicting postoperative ICU admission.

OBJECTIVE

To determine if the SURPAS model could accurately predict postoperative ICU admission in a broad surgical population.

DESIGN, SETTING, AND PARTICIPANTS

This decision analytical model was a retrospective, observational analysis of prospectively collected patient data from the 2012 to 2018 American College of Surgeons (ACS) National Surgical Quality Improvement Program (NSQIP) database, which were merged with individual patients' electronic health record data to capture postoperative ICU use. Multivariable logistic regression modeling was used to determine how the 8 preoperative variables of the SURPAS model predicted ICU use compared with a model inputting all 28 preoperatively available NSQIP variables. Data included in the analysis were collected for the ACS NSQIP at 5 hospitals (1 tertiary academic center, 4 academic affiliated hospitals) within the University of Colorado Health System between January 1, 2012, and December 31, 2018. Included patients were those undergoing surgery in 9 surgical specialties during the 2012 to 2018 period. Data were analyzed from May 29 to July 30, 2021.

EXPOSURE

Surgery in 9 surgical specialties, including general, gynecology, orthopedic, otolaryngology, plastic, thoracic, urology, vascular, and neurosurgery.

MAIN OUTCOMES AND MEASURES

Use of ICU care up to 30 days after surgery.

RESULTS

A total of 34 568 patients were included in the analytical data set: 32 032 (92.7%) in the cohort without postoperative ICU use and 2545 (7.4%) in the cohort with postoperative ICU use (no ICU use: mean [SD] age, 54.9 [16.6] years; 18 188 women [56.8%]; ICU use: mean [SD] age, 60.3 [15.3] years; 1333 men [52.4%]). For the internal chronologic validation of the 7-variable SURPAS model, data from 2012 to 2016 were used as the training data set (n = 24 250, 70.2% of the total sample size of 34 568) and data from 2017 to 2018 were used as the test data set (n = 10 318, 29.8% of the total sample size of 34 568). The C statistic improved in the test data set compared with the training data set (0.933; 95% CI, 0.924-0.941 vs 0.922; 95% CI, 0.917-0.928), whereas the Brier score was slightly worse in the test data set compared with the training data set (0.045; 95% CI, 0.042-0.048 vs 0.045; 95% CI, 0.043-0.047). The SURPAS model compared favorably with the model inputting all 28 NSQIP variables, with both having good calibration between observed and expected outcomes in the Hosmer-Lemeshow graphs and similar Brier scores (model inputting all variables, 0.044; 95% CI, 0.043-0.048; SURPAS model, 0.045; 95% CI, 0.042-0.046) and C statistics (model inputting all variables, 0.929; 95% CI, 0.925-0.934; SURPAS model, 0.925; 95% CI, 0.921-0.930).

CONCLUSIONS AND RELEVANCE

Results of this decision analytical model study revealed that the SURPAS prediction model accurately predicted postoperative ICU use across a diverse surgical population. These results suggest that the SURPAS prediction model can be used to help with preoperative planning and resource allocation of limited ICU beds.

Artificial intelligence in perioperative medicine - a narrative review

Recent advancements in artificial intelligence (AI) techniques have enabled the development of accurate prediction models using clinical big data. AI models for perioperative risk stratification, intraoperative event prediction, biosignal analyses, and intensive care medicine have been developed in the field of perioperative medicine. Some of these models have been validated using external datasets and randomized controlled trials. Once these models are implemented in electronic health record systems or software medical devices, they could help anesthesiologists improve clinical outcomes by accurately predicting complications and suggesting optimal treatment strategies in real-time. This review provides an overview of the AI techniques used in perioperative medicine and a summary of the studies that have been published using these techniques. Understanding these techniques will aid in their appropriate application in clinical practice.

Machine learning to guide clinical decision-making in abdominal surgery-a systematic literature review

PURPOSE

An indication for surgical therapy includes balancing benefits against risk, which remains a key task in all surgical disciplines. Decisions are oftentimes based on clinical experience while guidelines lack evidence-based background. Various medical fields capitalized the application of machine learning (ML), and preliminary research suggests promising implications in surgeons' workflow. Hence, we evaluated ML's contemporary and possible future role in clinical decision-making (CDM) focusing on abdominal surgery.

METHODS

Using the PICO framework, relevant keywords and research questions were identified. Following the PRISMA guidelines, a systemic search strategy in the PubMed database was conducted. Results were filtered by distinct criteria and selected articles were manually full text reviewed.

RESULTS

Literature review revealed 4,396 articles, of which 47 matched the search criteria. The mean number of patients included was 55,843. A total of eight distinct ML techniques were evaluated whereas AUROC was applied by most authors for comparing ML predictions vs. conventional CDM routines. Most authors (N = 30/47, 63.8%) stated ML's superiority in the prediction of benefits and risks of surgery. The identification of highly relevant parameters to be integrated into algorithms allowing a more precise prognosis was emphasized as the main advantage of ML in CDM.

CONCLUSIONS

A potential value of ML for surgical decision-making was demonstrated in several scientific articles. However, the low number of publications with only few collaborative studies between surgeons and computer scientists underpins the early phase of this highly promising field. Interdisciplinary research initiatives combining existing clinical datasets and emerging techniques of data processing may likely improve CDM in abdominal surgery in the future.

Using machine learning to classify patients on opioid use

Objectives

High-frequent opioid use tends to increase an individual's risk of opioid use disorder, overdose and death. Thus, it is important to predict an individuals' opioid use frequency to improve opioid prescription utilization outcomes.

Methods

Individuals receiving at least one opioid prescription from 2016 to 2018 in the national representative data, Medical Expenditure Panel Survey, were included. This study applied five machine learning (ML) techniques, including support vector machine, random forest, neural network, gradient boosting and XGBoost (extreme gradient boosting), to predict opioid use frequency. This study compared the performance of these ML models with penalized logistic regression. The study outcome was whether an individual lied in the upper 10% of the opioid prescription distribution. Predictors were selected based on Gelberg-Andersen's Behavioral Model of Health Services Utilization. The prediction performance was assessed using the area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC) in the test data. Patient characteristics as predictors for high-frequency use of opioids were ranked by the relative importance in prediction in the test data.

Key findings

Random forest and gradient boosting achieved the top values of both AUROC and AUPRC, outperforming logistic regression and three other ML methods. In the best performing model, the random forest, the following characteristics had high predictive power in the frequency of opioid use: age, number of chronic conditions, public insurance and self-perceived health status.

Conclusions

The results of this study demonstrate that ML techniques can be a promising and powerful technique in predicting the frequency of opioid use and health outcomes.

Machine learning classifiers do not improve prediction of hospitalization > 2 days after fast-track hip and knee arthroplasty compared with a classical statistical risk model

Background and purpose: Prediction of postoperative outcomes and length of hospital stay (LOS) of patients is vital for allocation of healthcare resources. We investigated the performance of prediction models based on machinelearning algorithms compared with a previous risk stratification model using traditional multiple logistic regression, for predicting the risk of a LOS of > 2 days after fast-track total hip and knee replacement. Patients and methods: 3 different machine learning classifiers were trained on data from the Lundbeck Centre for Fast-track Hip and Knee Replacement Database (LCDB) collected from 9,512 patients between 2016 and 2017. The chosen classifiers were a random forest classifier (RF), a support vector machine classifier with a polynomial kernel (SVM), and a multinomial Naïve-Bayes classifier (NB). Results: Comparing performance measures of the classifiers with the traditional model revealed that all the models had a similar performance in terms of F1 score, accuracy, sensitivity, specificity, area under the receiver operating curve (AUC), and area under the precision-recall curve (AUPRC). A feature importance analysis of the RF classifier found hospital, age, use of walking aid, living alone, and joint operated on to be the most relevant input features. None of the classifiers reached a clinically relevant performance with the input data from the LCDB. Interpretation: Despite the promising prospects of machine-learning practices for disease and risk prediction, none of the machine learning models tested outperformed the traditional multiple regression model in predicting which patients in this cohort had a LOS > 2 days.

Machine learning in perioperative medicine: a systematic review

Background

Risk stratification plays a central role in anesthetic evaluation. The use of Big Data and machine learning (ML) offers considerable advantages for collection and evaluation of large amounts of complex health-care data. We conducted a systematic review to understand the role of ML in the development of predictive post-surgical outcome models and risk stratification.

Methods

Following the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines, we selected the period of the research for studies from 1 January 2015 up to 30 March 2021. A systematic search in Scopus, CINAHL, the Cochrane Library, PubMed, and MeSH databases was performed; the strings of research included different combinations of keywords: “risk prediction,” “surgery,” “machine learning,” “intensive care unit (ICU),” and “anesthesia” “perioperative.” We identified 36 eligible studies. This study evaluates the quality of reporting of prediction models using the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) checklist.

Results

The most considered outcomes were mortality risk, systemic complications (pulmonary, cardiovascular, acute kidney injury (AKI), etc.), ICU admission, anesthesiologic risk and prolonged length of hospital stay. Not all the study completely followed the TRIPOD checklist, but the quality was overall acceptable with 75% of studies (Rev #2, comm #minor issue) showing an adherence rate to TRIPOD more than 60%. The most frequently used algorithms were gradient boosting (n = 13), random forest (n = 10), logistic regression (LR; n = 7), artificial neural networks (ANNs; n = 6), and support vector machines (SVM; n = 6). Models with best performance were random forest and gradient boosting, with AUC > 0.90.

Conclusions

The application of ML in medicine appears to have a great potential. From our analysis, depending on the input features considered and on the specific prediction task, ML algorithms seem effective in outcomes prediction more accurately than validated prognostic scores and traditional statistics. Thus, our review encourages the healthcare domain and artificial intelligence (AI) developers to adopt an interdisciplinary and systemic approach to evaluate the overall impact of AI on perioperative risk assessment and on further health care settings as well.

Impact of Intraoperative Data on Risk Prediction for Mortality After Intra-Abdominal Surgery

BACKGROUND

Risk prediction models for postoperative mortality after intra-abdominal surgery have typically been developed using preoperative variables. It is unclear if intraoperative data add significant value to these risk prediction models.

METHODS

With IRB approval, an institutional retrospective cohort of intra-abdominal surgery patients in the 2005 to 2015 American College of Surgeons National Surgical Quality Improvement Program was identified. Intraoperative data were obtained from the electronic health record. The primary outcome was 30-day mortality. We evaluated the performance of machine learning algorithms to predict 30-day mortality using: 1) baseline variables and 2) baseline + intraoperative variables. Algorithms evaluated were: 1) logistic regression with elastic net selection, 2) random forest (RF), 3) gradient boosting machine (GBM), 4) support vector machine (SVM), and 5) convolutional neural networks (CNNs). Model performance was evaluated using the area under the receiver operator characteristic curve (AUROC). The sample was randomly divided into a training/testing split with 80%/20% probabilities. Repeated 10-fold cross-validation identified the optimal model hyperparameters in the training dataset for each model, which were then applied to the entire training dataset to train the model. Trained models were applied to the test cohort to evaluate model performance. Statistical significance was evaluated using P < .05.

RESULTS

The training and testing cohorts contained 4322 and 1079 patients, respectively, with 62 (1.4%) and 15 (1.4%) experiencing 30-day mortality, respectively. When using only baseline variables to predict mortality, all algorithms except SVM (area under the receiver operator characteristic curve [AUROC], 0.83 [95% confidence interval {CI}, 0.69-0.97]) had AUROC >0.9: GBM (AUROC, 0.96 [0.94-1.0]), RF (AUROC, 0.96 [0.92-1.0]), CNN (AUROC, 0.96 [0.92-0.99]), and logistic regression (AUROC, 0.95 [0.91-0.99]). AUROC significantly increased with intraoperative variables with CNN (AUROC, 0.97 [0.96-0.99]; P = .047 versus baseline), but there was no improvement with GBM (AUROC, 0.97 [0.95-0.99]; P = .3 versus baseline), RF (AUROC, 0.96 [0.93-1.0]; P = .5 versus baseline), and logistic regression (AUROC, 0.94 [0.90-0.99]; P = .6 versus baseline).

CONCLUSIONS

Postoperative mortality is predicted with excellent discrimination in intra-abdominal surgery patients using only preoperative variables in various machine learning algorithms. The addition of intraoperative data to preoperative data also resulted in models with excellent discrimination, but model performance did not improve.

Machine Learning Approach Using Routine Immediate Postoperative Laboratory Values for Predicting Postoperative Mortality

Background: Several prediction models have been proposed for preoperative risk stratification for mortality. However, few studies have investigated postoperative risk factors, which have a significant influence on survival after surgery. This study aimed to develop prediction models using routine immediate postoperative laboratory values for predicting postoperative mortality. Methods: Two tertiary hospital databases were used in this research: one for model development and another for external validation of the resulting models. The following algorithms were utilized for model development: LASSO logistic regression, random forest, deep neural network, and XGBoost. We built the models on the lab values from immediate postoperative blood tests and compared them with the SASA scoring system to demonstrate their efficacy. Results: There were 3817 patients who had immediate postoperative blood test values. All models trained on immediate postoperative lab values outperformed the SASA model. Furthermore, the developed random forest model had the best AUROC of 0.82 and AUPRC of 0.13, and the phosphorus level contributed the most to the random forest model. Conclusions: Machine learning models trained on routine immediate postoperative laboratory values outperformed previously published approaches in predicting 30-day postoperative mortality, indicating that they may be beneficial in identifying patients at increased risk of postoperative death.

Perioperative Risk Assessment in Pancreatic Surgery Using Machine Learning

Pancreatic surgery is associated with a high risk for postoperative complications and death of patients. Complications occur in a variable interval after the procedure. Often, a patient has already left the ICU and is not properly monitored anymore when the complication occurs. Risk stratification models can assist in identifying patients at risk in order to keep these patients in ICU for longer. This, in turn, helps to identify complications earlier and increase survival rates. We trained multiple machine learning models on pre-, intra- and short term postoperative data from patients who underwent pancreatic resection at the Department of Surgery, Campus Charité Mitte | Campus Virchow-Klinikum, Charité - Universitätsmedizin Berlin. The presented models achieve an area under the precision-recall curve (AUPRC) of up to 0.51 for predicting patient death and 0.53 for predicting a specific major complication. Overall, we found that a classical logistic regression model performs best for the investigated classification tasks. As more patient data becomes available throughout the perioperative stay, the performance of the risk stratification model improves and should therefore repeatedly be computed.

Artificial Intelligence Surgery: How Do We Get to Autonomous Actions in Surgery?

Most surgeons are skeptical as to the feasibility of autonomous actions in surgery. Interestingly, many examples of autonomous actions already exist and have been around for years. Since the beginning of this millennium, the field of artificial intelligence (AI) has grown exponentially with the development of machine learning (ML), deep learning (DL), computer vision (CV) and natural language processing (NLP). All of these facets of AI will be fundamental to the development of more autonomous actions in surgery, unfortunately, only a limited number of surgeons have or seek expertise in this rapidly evolving field. As opposed to AI in medicine, AI surgery (AIS) involves autonomous movements. Fortuitously, as the field of robotics in surgery has improved, more surgeons are becoming interested in technology and the potential of autonomous actions in procedures such as interventional radiology, endoscopy and surgery. The lack of haptics, or the sensation of touch, has hindered the wider adoption of robotics by many surgeons; however, now that the true potential of robotics can be comprehended, the embracing of AI by the surgical community is more important than ever before. Although current complete surgical systems are mainly only examples of tele-manipulation, for surgeons to get to more autonomously functioning robots, haptics is perhaps not the most important aspect. If the goal is for robots to ultimately become more and more independent, perhaps research should not focus on the concept of haptics as it is perceived by humans, and the focus should be on haptics as it is perceived by robots/computers. This article will discuss aspects of ML, DL, CV and NLP as they pertain to the modern practice of surgery, with a focus on current AI issues and advances that will enable us to get to more autonomous actions in surgery. Ultimately, there may be a paradigm shift that needs to occur in the surgical community as more surgeons with expertise in AI may be needed to fully unlock the potential of AIS in a safe, efficacious and timely manner.

Optimizing predictive strategies for acute kidney injury after major vascular surgery

BACKGROUND

Postoperative acute kidney injury is common after major vascular surgery and is associated with increased morbidity, mortality, and cost. High-performance risk stratification using a machine learning model can inform strategies that mitigate harm and optimize resource use. It is hypothesized that incorporating intraoperative data would improve machine learning model accuracy, discrimination, and precision in predicting acute kidney injury among patients undergoing major vascular surgery.

METHODS

A single-center retrospective cohort of 1,531 adult patients who underwent nonemergency major vascular surgery, including open aortic, endovascular aortic, and lower extremity bypass procedures, was evaluated. The validated, automated MySurgeryRisk analytics platform used electronic health record data to forecast patient-level probabilistic risk scores for postoperative acute kidney injury using random forest models with preoperative data alone and perioperative data (preoperative plus intraoperative). The MySurgeryRisk predictions were compared with each other as well as with the American Society of Anesthesiologists physical status classification.

RESULTS

Machine learning models using perioperative data had greater accuracy, discrimination, and precision than models using either preoperative data alone or the American Society of Anesthesiologists physical status classification (accuracy: 0.70 vs 0.64 vs 0.62, area under the receiver operating characteristics curve: 0.77 vs 0.68 vs 0.61, area under the precision-recall curve: 0.70 vs 0.58 vs 0.48).

CONCLUSION

In predicting acute kidney injury after major vascular surgery, machine learning approaches that incorporate dynamic intraoperative data had greater accuracy, discrimination, and precision than models using either preoperative data alone or the American Society of Anesthesiologists physical status classification. Machine learning methods have the potential for real-time identification of high-risk patients who may benefit from personalized risk-reduction strategies.

American Society of Anesthesiologists Physical Status Classification as a reliable predictor of postoperative medical complications and mortality following ambulatory surgery: an analysis of 2,089,830 ACS-NSQIP outpatient cases

BACKGROUND

Seventy percent of surgical procedures are currently performed in the outpatient setting. Although the American Society of Anesthesiologists (ASA) Physical Classification ability to predict risk has been evaluated for in-patient surgeries, an evaluation in outpatient surgeries has yet to be performed. The major goal of the current study is to determine if the ASA classification is an independent predictor for morbidity and mortality for outpatient surgeries.

METHODS

The 2005 through 2016 NSQIP Participant Use Data Files were queried to extract all patients scheduled for outpatient surgery. ASA PS class was the primary independent variable of interest. The primary outcome was 30-day medical complications, defined as having one or more of the following postoperative outcomes: (1) deep vein thrombosis, (2) pulmonary embolism, (3) reintubation, (4) failure to wean from ventilator, (5) renal insufficiency, (6) renal failure, (7) stroke, (8) cardiac arrest, (9) myocardial infarction, (10) pneumonia, (11) urinary tract infection, (12) systemic sepsis or septic shock. Mortality was also evaluated as a separate outcome.

RESULTS

A total of 2,089,830 cases were included in the study. 24,777 (1.19%) patients had medical complications and 1,701 (0.08%) died within 30 days. ASA PS IV patients had a much greater chance of dying when compared to healthy patients, OR (95%CI) of 89 (55 to 143), P < 0.001. Nonetheless, over 30,000 ASA PS IV patients had surgery in the outpatient setting. Multivariable analysis demonstrated a stepwise independent association between ASA PS class and medical complications (C statistic = 0.70), mortality (C statistic = 0.74) and readmissions (C statistic = 0.67). Risk stratifying ability was maintained across surgical procedures and anesthesia techniques.

CONCLUSIONS

ASA PS class is a simple risk stratification tool for surgeries in the outpatient setting. Patients with higher ASA PS classes subsequently developed medical complications or mortality at a greater frequency than patients with lower ASA PS class after outpatient surgery. Our results suggest that the ambulatory setting may not be able to match the needs of high-risk patients.