Interpretable artificial intelligence to optimise use of imatinib after resection in patients with localised gastrointestinal stromal tumours: an observational cohort study.

BACKGROUND: Current guidelines recommend use of adjuvant imatinib therapy for many patients with gastrointestinal stromal tumours (GISTs); however, its optimal treatment duration is unknown and some patient groups do not benefit from the therapy. We aimed to apply state-of-the-art, interpretable artificial intelligence (ie, predictions or prescription logic that can be easily understood) methods on real-world data to establish which groups of patients with GISTs should receive adjuvant imatinib, its optimal treatment duration, and the benefits conferred by this therapy. METHODS: In this observational cohort study, we considered for inclusion all patients who underwent resection of primary, non-metastatic GISTs at the Memorial Sloan Kettering Cancer Center (MSKCC; New York, NY, USA) between Oct 1, 1982, and Dec 31, 2017, and who were classified as intermediate or high risk according to the Armed Forces Institute of Pathology Miettinen criteria and had complete follow-up data with no missing entries. A counterfactual random forest model, which used predictors of recurrence (mitotic count, tumour size, and tumour site) and imatinib duration to infer the probability of recurrence at 7 years for a given patient under each duration of imatinib treatment, was trained in the MSKCC cohort. Optimal policy trees (OPTs), a state-of-the-art interpretable AI-based method, were used to read the counterfactual random forest model by training a decision tree with the counterfactual predictions. The OPT recommendations were externally validated in two cohorts of patients from Poland (the Polish Clinical GIST Registry), who underwent GIST resection between Dec 1, 1981, and Dec 31, 2011, and from Spain (the Spanish Group for Research in Sarcomas), who underwent resection between Oct 1, 1987, and Jan 30, 2011. FINDINGS: Among 1007 patients who underwent GIST surgery in MSKCC, 117 were included in the internal cohort; for the external cohorts, the Polish cohort comprised 363 patients and the Spanish cohort comprised 239 patients. The OPT did not recommend imatinib for patients with GISTs of gastric origin measuring less than 15·9 cm with a mitotic count of less than 11·5 mitoses per 5 mm2 or for those with small GISTs (<5·4 cm) of any site with a count of less than 11·5 mitoses per 5 mm2. In this cohort, the OPT cutoffs had a sensitivity of 92·7% (95% CI 82·4-98·0) and a specificity of 33·9% (22·3-47·0). The application of these cutoffs in the two external cohorts would have spared 38 (29%) of 131 patients in the Spanish cohort and 44 (35%) of 126 patients in the Polish cohort from unnecessary treatment with imatinib. Meanwhile, the risk of undertreating patients in these cohorts was minimal (sensitivity 95·4% [95% CI 89·5-98·5] in the Spanish cohort and 92·4% [88·3-95·4] in the Polish cohort). The OPT tested 33 different durations of imatinib treatment (<5 years) and found that 5 years of treatment conferred the most benefit. INTERPRETATION: If the identified patient subgroups were applied in clinical practice, as many as a third of the current cohort of candidates who do not benefit from adjuvant imatinib would be encouraged to not receive imatinib, subsequently avoiding unnecessary toxicity on patients and financial strain on health-care systems. Our finding that 5 years is the optimal duration of imatinib treatment could be the best source of evidence to inform clinical practice until 2028, when a randomised controlled trial with the same aims is expected to report its findings. FUNDING: National Cancer Institute.

Validation of the Al-based Predictive OpTimal Trees in Emergency Surgery Risk (POTTER) Calculator in Patients 65 Years and Older.

OBJECTIVE: We sought to assess the performance of the Predictive OpTimal Trees in Emergency Surgery Risk (POTTER) tool in elderly emergency surgery (ES) patients. SUMMARY BACKGROUND DATA: The POTTER tool was derived using a novel Artificial Intelligence (AI)-methodology called optimal classification trees and validated for prediction of ES outcomes. POTTER outperforms all existent risk-prediction models and is available as an interactive smartphone application. Predicting outcomes in elderly patients has been historically challenging and POTTER has not yet been tested in this population. METHODS: All patients ≥65 years who underwent ES in the ACS-NSQIP 2017 database were included. POTTER's performance for 30-day mortality and 18 postoperative complications (eg, respiratory or renal failure) was assessed using c-statistic methodology, with planned sub-analyses for patients 65 to 74, 75 to 84, and 85+ years. RESULTS: A total of 29,366 patients were included, with mean age 77, 55.8% females, and 62% who underwent emergency general surgery. POTTER predicted mortality accurately in all patients over 65 (c-statistic 0.80). Its best performance was in patients 65 to 74 years (c-statistic 0.84), and its worst in patients ≥85 years (c-statistic 0.71). POTTER had the best discrimination for predicting septic shock (c-statistic 0.90), respiratory failure requiring mechanical ventilation for ≥48 hours (c-statistic 0.86), and acute renal failure (c-statistic 0.85). CONCLUSIONS: POTTER is a novel, interpretable, and highly accurate predictor of in-hospital mortality in elderly ES patients up to age 85 years. POTTER could prove useful for bedside counseling and for benchmarking of ES care.

Optimizing schools' start time and bus routes.

Maintaining a fleet of buses to transport students to school is a major expense for school districts. To reduce costs by reusing buses between schools, many districts spread start times across the morning. However, assigning each school a time involves estimating the impact on transportation costs and reconciling additional competing objectives. Facing this intricate optimization problem, school districts must resort to ad hoc approaches, which can be expensive, inequitable, and even detrimental to student health. For example, there is medical evidence that early high school starts are impacting the development of an entire generation of students and constitute a major public health crisis. We present an optimization model for the school time selection problem (STSP), which relies on a school bus routing algorithm that we call biobjective routing decomposition (BiRD). BiRD leverages a natural decomposition of the routing problem, computing and combining subproblem solutions via mixed integer optimization. It significantly outperforms state-of-the-art routing methods, and its implementation in Boston has led to $5 million in yearly savings, maintaining service quality for students despite a 50-bus fleet reduction. Using BiRD, we construct a tractable proxy to transportation costs, allowing the formulation of the STSP as a multiobjective generalized quadratic assignment problem. Local search methods provide high-quality solutions, allowing school districts to explore tradeoffs between competing priorities and choose times that best fulfill community needs. In December 2017, the development of this method led the Boston School Committee to unanimously approve the first school start time reform in 30 years.

Machine learning models for mitral valve replacement: A comparative analysis with the Society of Thoracic Surgeons risk score.

BACKGROUND: Current Society of Thoracic Surgeons (STS) risk models for predicting outcomes of mitral valve surgery (MVS) assume a linear and cumulative impact of variables. We evaluated postoperative MVS outcomes and designed mortality and morbidity risk calculators to supplement the STS risk score. METHODS: Data from the STS Adult Cardiac Surgery Database for MVS was used from 2008 to 2017. The data included 383,550 procedures and 89 variables. Machine learning (ML) algorithms were employed to train models to predict postoperative outcomes for MVS patients. Each model's discrimination and calibration performance were validated using unseen data against the STS risk score. RESULTS: Comprehensive mortality and morbidity risk assessment scores were derived from a training set of 287,662 observations. The area under the curve (AUC) for mortality ranged from 0.77 to 0.83, leading to a 3% increase in predictive accuracy compared to the STS score. Logistic Regression and eXtreme Gradient Boosting achieved the highest AUC for prolonged ventilation (0.82) and deep sternal wound infection (0.78 and 0.77) respectively. EXtreme Gradient Boosting performed the best with an AUC of 0.815 for renal failure. For permanent stroke prediction all models performed similarly with an AUC around 0.67. The ML models led to improved calibration performance for mortality, prolonged ventilation, and renal failure, especially in cases of reconstruction/repair and replacement surgery. CONCLUSIONS: The proposed risk models complement existing STS models in predicting mortality, prolonged ventilation, and renal failure, allowing healthcare providers to more accurately assess a patient's risk of morbidity and mortality when undergoing MVS.

Selecting Children with Vesicoureteral Reflux Who are Most Likely to Benefit from Antibiotic Prophylaxis: Application of Machine Learning to RIVUR.

PURPOSE: Continuous antibiotic prophylaxis reduces the risk of recurrent urinary tract infection by 50% in children with vesicoureteral reflux. However, there may be subgroups in whom continuous antibiotic prophylaxis could be used more selectively. We sought to develop a machine learning model to identify such subgroups. MATERIALS AND METHODS: We used RIVUR data, randomly split into train/test in a 4:1 ratio. Two models were developed to predict recurrent urinary tract infection risk in scenario with and without continuous antibiotic prophylaxis. The test set was then used to validate recurrent urinary tract infection events and the effectiveness of continuous antibiotic prophylaxis. Predicted probabilities of recurrent urinary tract infection were generated from each model. Continuous antibiotic prophylaxis was assigned at various cutoffs of recurrent urinary tract infection risk reduction to evaluate continuous antibiotic prophylaxis effectiveness. RESULTS: A total of 607 patients (558 female/49 male, median age 12 months) were included. Predictors included vesicoureteral reflux grade, serum creatinine, race/gender, prior urinary tract infection symptoms (fever/dysuria) and weight percentiles. The AUC of the prediction model of recurrent urinary tract infection (continuous antibiotic prophylaxis/placebo) was 0.82 (95% CI 0.74-0.87). Using 10% recurrent urinary tract infection risk reduction cutoff, minimal recurrent urinary tract infection per population level can be achieved by giving continuous antibiotic prophylaxis to 40% of patients with vesicoureteral reflux instead of everyone. In a test set (121), 51 patients had continuous antibiotic prophylaxis randomization consistent with model recommendation (continuous antibiotic prophylaxis if recurrent urinary tract infection risk reduction >10%). Recurrent urinary tract infection incidence was significantly lower among this group compared to those whose continuous antibiotic prophylaxis assignment differed from model suggestion (7.5% vs 19.4%, p=0.037). CONCLUSIONS: Our predictive model identifies patients with vesicoureteral reflux who are more likely to benefit from continuous antibiotic prophylaxis, which would allow more selective, personalized use of continuous antibiotic prophylaxis with maximal benefit, while minimizing use in those with least need.

Personalized prescription of ACEI/ARBs for hypertensive COVID-19 patients.

The COVID-19 pandemic has prompted an international effort to develop and repurpose medications and procedures to effectively combat the disease. Several groups have focused on the potential treatment utility of angiotensin-converting-enzyme inhibitors (ACEIs) and angiotensin-receptor blockers (ARBs) for hypertensive COVID-19 patients, with inconclusive evidence thus far. We couple electronic medical record (EMR) and registry data of 3,643 patients from Spain, Italy, Germany, Ecuador, and the US with a machine learning framework to personalize the prescription of ACEIs and ARBs to hypertensive COVID-19 patients. Our approach leverages clinical and demographic information to identify hospitalized individuals whose probability of mortality or morbidity can decrease by prescribing this class of drugs. In particular, the algorithm proposes increasing ACEI/ARBs prescriptions for patients with cardiovascular disease and decreasing prescriptions for those with low oxygen saturation at admission. We show that personalized recommendations can improve patient outcomes by 1.0% compared to the standard of care when applied to external populations. We develop an interactive interface for our algorithm, providing physicians with an actionable tool to easily assess treatment alternatives and inform clinical decisions. This work offers the first personalized recommendation system to accurately evaluate the efficacy and risks of prescribing ACEIs and ARBs to hypertensive COVID-19 patients.

From predictions to prescriptions: A data-driven response to COVID-19.

The COVID-19 pandemic has created unprecedented challenges worldwide. Strained healthcare providers make difficult decisions on patient triage, treatment and care management on a daily basis. Policy makers have imposed social distancing measures to slow the disease, at a steep economic price. We design analytical tools to support these decisions and combat the pandemic. Specifically, we propose a comprehensive data-driven approach to understand the clinical characteristics of COVID-19, predict its mortality, forecast its evolution, and ultimately alleviate its impact. By leveraging cohort-level clinical data, patient-level hospital data, and census-level epidemiological data, we develop an integrated four-step approach, combining descriptive, predictive and prescriptive analytics. First, we aggregate hundreds of clinical studies into the most comprehensive database on COVID-19 to paint a new macroscopic picture of the disease. Second, we build personalized calculators to predict the risk of infection and mortality as a function of demographics, symptoms, comorbidities, and lab values. Third, we develop a novel epidemiological model to project the pandemic's spread and inform social distancing policies. Fourth, we propose an optimization model to re-allocate ventilators and alleviate shortages. Our results have been used at the clinical level by several hospitals to triage patients, guide care management, plan ICU capacity, and re-distribute ventilators. At the policy level, they are currently supporting safe back-to-work policies at a major institution and vaccine trial location planning at Janssen Pharmaceuticals, and have been integrated into the US Center for Disease Control's pandemic forecast.

Personalized treatment for coronary artery disease patients: a machine learning approach.

Current clinical practice guidelines for managing Coronary Artery Disease (CAD) account for general cardiovascular risk factors. However, they do not present a framework that considers personalized patient-specific characteristics. Using the electronic health records of 21,460 patients, we created data-driven models for personalized CAD management that significantly improve health outcomes relative to the standard of care. We develop binary classifiers to detect whether a patient will experience an adverse event due to CAD within a 10-year time frame. Combining the patients' medical history and clinical examination results, we achieve 81.5% AUC. For each treatment, we also create a series of regression models that are based on different supervised machine learning algorithms. We are able to estimate with average R2 = 0.801 the outcome of interest; the time from diagnosis to a potential adverse event (TAE). Leveraging combinations of these models, we present ML4CAD, a novel personalized prescriptive algorithm. Considering the recommendations of multiple predictive models at once, the goal of ML4CAD is to identify for every patient the therapy with the best expected TAE using a voting mechanism. We evaluate its performance by measuring the prescription effectiveness and robustness under alternative ground truths. We show that our methodology improves the expected TAE upon the current baseline by 24.11%, increasing it from 4.56 to 5.66 years. The algorithm performs particularly well for the male (24.3% improvement) and Hispanic (58.41% improvement) subpopulations. Finally, we create an interactive interface, providing physicians with an intuitive, accurate, readily implementable, and effective tool.

The rise and fall of the model for end-stage liver disease score and the need for an optimized machine learning approach for liver allocation.

PURPOSE OF REVIEW: The Model for End-Stage Liver Disease (MELD) has been used to rank liver transplant candidates since 2002, and at the time bringing much needed objectivity to the liver allocation process. However, and despite numerous revisions to the MELD score, current liver allocation still does not allow for equitable access to all waitlisted liver candidates. RECENT FINDINGS: An optimized prediction of mortality (OPOM) was developed utilizing novel machine-learning optimal classification tree models trained to predict a liver candidate's 3-month waitlist mortality or removal. When compared to MELD and MELD-Na, OPOM more accurately and objectively prioritized candidates for liver transplantation based on disease severity. In simulation analysis, OPOM allowed for more equitable allocation of livers with a resultant significant number of additional lives saved every year when compared with MELD-based allocation. SUMMARY: Machine learning technology holds the potential to help guide transplant clinical practice, and thus potentially guide national organ allocation policy.

Development and validation of an optimized prediction of mortality for candidates awaiting liver transplantation.

Since 2002, the Model for End-Stage Liver Disease (MELD) has been used to rank liver transplant candidates. However, despite numerous revisions, MELD allocation still does not allow for equitable access to all waitlisted candidates. An optimized prediction of mortality (OPOM) was developed (http://www.opom.online) utilizing machine-learning optimal classification tree models trained to predict a candidate's 3-month waitlist mortality or removal utilizing the Standard Transplant Analysis and Research (STAR) dataset. The Liver Simulated Allocation Model (LSAM) was then used to compare OPOM to MELD-based allocation. Out-of-sample area under the curve (AUC) was also calculated for candidate groups of increasing disease severity. OPOM allocation, when compared to MELD, reduced mortality on average by 417.96 (406.8-428.4) deaths every year in LSAM analysis. Improved survival was noted across all candidate demographics, diagnoses, and geographic regions. OPOM delivered a substantially higher AUC across all disease severity groups. OPOM more accurately and objectively prioritizes candidates for liver transplantation based on disease severity, allowing for more equitable allocation of livers with a resultant significant number of additional lives saved every year. These data demonstrate the potential of machine learning technology to help guide clinical practice, and potentially guide national policy.

Clinical benefit, toxicity and cost of metastatic breast cancer therapies: systematic review and meta-analysis.

PURPOSE: Oncologists, clinical trialists, and guideline developers need tools that enable them to efficiently review the settings and results of previous studies testing metastatic breast cancer (MBC) drug therapies. METHODS: We searched the literature to identify clinical trials testing MBC drug therapies. Key eligibility criteria included at least 90% of patients enrolled in the trial having MBC, therapeutic clinical trials, and Phase II-III studies. Studies were stratified based on patients' tumor receptor statuses and prior exposure to therapy. Survival and toxicity of each drug therapy were estimated from randomized controlled trials using network meta-analysis and from all studies using meta-analysis. These results, along with estimated drug costs, are presented in a web-based visualization tool. RESULTS: We included 1865 studies containing 2676 treatment arms and 184,563 patients in the tool ( http://www.cancertrials.info ). Meta-analysis-based efficacy and toxicity estimates are available for 85 HER-2-directed therapies, 84 hormonal therapies, and 442 undirected therapies. Network meta-analysis-based estimates are available for 16 HER-2-directed therapies, 26 hormonal therapies, and 131 undirected therapies. CONCLUSIONS: In this era of increasing choices of MBC therapeutic agents and no superior approach to choosing a treatment regimen, the ability to compare multiple therapies based on survival, toxicity and cost would enable treating physicians to optimize therapeutic choices for patients. For investigators, it can point them in research directions that were previously non-obvious and for guideline designers, enable them to efficiently review the MBC clinical trial literature and visualize how regimens compare in the key dimensions of clinical benefit, toxicity, and cost.

Surgical Risk Is Not Linear: Derivation and Validation of a Novel, User-friendly, and Machine-learning-based Predictive OpTimal Trees in Emergency Surgery Risk (POTTER) Calculator.

INTRODUCTION: Most risk assessment tools assume that the impact of risk factors is linear and cumulative. Using novel machine-learning techniques, we sought to design an interactive, nonlinear risk calculator for Emergency Surgery (ES). METHODS: All ES patients in the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) 2007 to 2013 database were included (derivation cohort). Optimal Classification Trees (OCT) were leveraged to train machine-learning algorithms to predict postoperative mortality, morbidity, and 18 specific complications (eg, sepsis, surgical site infection). Unlike classic heuristics (eg, logistic regression), OCT is adaptive and reboots itself with each variable, thus accounting for nonlinear interactions among variables. An application [Predictive OpTimal Trees in Emergency Surgery Risk (POTTER)] was then designed as the algorithms' interactive and user-friendly interface. POTTER performance was measured (c-statistic) using the 2014 ACS-NSQIP database (validation cohort) and compared with the American Society of Anesthesiologists (ASA), Emergency Surgery Score (ESS), and ACS-NSQIP calculators' performance. RESULTS: Based on 382,960 ES patients, comprehensive decision-making algorithms were derived, and POTTER was created where the provider's answer to a question interactively dictates the subsequent question. For any specific patient, the number of questions needed to predict mortality ranged from 4 to 11. The mortality c-statistic was 0.9162, higher than ASA (0.8743), ESS (0.8910), and ACS (0.8975). The morbidity c-statistics was similarly the highest (0.8414). CONCLUSION: POTTER is a highly accurate and user-friendly ES risk calculator with the potential to continuously improve accuracy with ongoing machine-learning. POTTER might prove useful as a tool for bedside preoperative counseling of ES patients and families.

Optimal healthcare decision making under multiple mathematical models: application in prostate cancer screening.

Important decisions related to human health, such as screening strategies for cancer, need to be made without a satisfactory understanding of the underlying biological and other processes. Rather, they are often informed by mathematical models that approximate reality. Often multiple models have been made to study the same phenomenon, which may lead to conflicting decisions. It is natural to seek a decision making process that identifies decisions that all models find to be effective, and we propose such a framework in this work. We apply the framework in prostate cancer screening to identify prostate-specific antigen (PSA)-based strategies that perform well under all considered models. We use heuristic search to identify strategies that trade off between optimizing the average across all models' assessments and being "conservative" by optimizing the most pessimistic model assessment. We identified three recently published mathematical models that can estimate quality-adjusted life expectancy (QALE) of PSA-based screening strategies and identified 64 strategies that trade off between maximizing the average and the most pessimistic model assessments. All prescribe PSA thresholds that increase with age, and 57 involve biennial screening. Strategies with higher assessments with the pessimistic model start screening later, stop screening earlier, and use higher PSA thresholds at earlier ages. The 64 strategies outperform 22 previously published expert-generated strategies. The 41 most "conservative" ones remained better than no screening with all models in extensive sensitivity analyses. We augment current comparative modeling approaches by identifying strategies that perform well under all models, for various degrees of decision makers' conservativeness.

Personalized Breast Cancer Screening.

PURPOSE: Abundant literature and clinical trials indicate that routine cancer screenings decrease patient mortality for several common cancers. However, current national cancer screening guidelines heavily rely on patient age as the predominant factor in deciding cancer screening timing, neglecting other important medical characteristics of individual patients. This approach either delays screening or prescribes excessive screenings. Another disadvantage of the current approach is its inability to combine information across hospital systems because of the lack of a coherent records system. METHODS: We propose to use claims data and medical insurance transactions that use consistent and pre-established sets of codes for diagnosis, procedures, and medications to develop a clinical support tool to supply supplemental insights and precautions for physicians to make more informed decisions. Furthermore, we propose a novel machine learning framework to recommend personalized, data-driven, and dynamic screening decisions. RESULTS: We apply this new method to the study of breast cancer mammograms using claims data from 378,840 female patients to demonstrate that across different risk populations, personalized screening reduces the average delay in a cancer diagnosis by 2-3 months with statistical significance, with even stronger benefits for individual patients up to 10 months. CONCLUSION: Incorporating personal medical characteristics using claims data and novel machine learning methodologies into breast cancer screening improves screening delay by more dynamically considering changing patient risks. Future incorporation of the proposed methodology in health care settings could be provided as a potential support tool for clinicians.

High-performance pediatric surgical risk calculator: A novel algorithm based on machine learning and pediatric NSQIP data.

BACKGROUNDS: New methods such as machine learning could provide accurate predictions with little statistical assumptions. We seek to develop prediction model of pediatric surgical complications based on pediatric National Surgical Quality Improvement Program(NSQIP). METHODS: All 2012-2018 pediatric-NSQIP procedures were reviewed. Primary outcome was defined as 30-day post-operative morbidity/mortality. Morbidity was further classified as any, major and minor. Models were developed using 2012-2017 data. 2018 data was used as independent performance evaluation. RESULTS: 431,148 patients were included in the 2012-2017 training and 108,604 were included in the 2018 testing set. Our prediction models had high performance in mortality prediction at 0.94 AUC in testing set. Our models outperformed ACS-NSQIP Calculator in all categories for morbidity (0.90 AUC for major, 0.86 AUC for any, 0.69 AUC in minor complications). CONCLUSIONS: We developed a high-performing pediatric surgical risk prediction model. This powerful tool could potentially be used to improve the surgical care quality.

Automated Segmentation of Sacral Chordoma and Surrounding Muscles Using Deep Learning Ensemble.

PURPOSE: The manual segmentation of organ structures in radiation oncology treatment planning is a time-consuming and highly skilled task, particularly when treating rare tumors like sacral chordomas. This study evaluates the performance of automated deep learning (DL) models in accurately segmenting the gross tumor volume (GTV) and surrounding muscle structures of sacral chordomas. METHODS AND MATERIALS: An expert radiation oncologist contoured 5 muscle structures (gluteus maximus, gluteus medius, gluteus minimus, paraspinal, piriformis) and sacral chordoma GTV on computed tomography images from 48 patients. We trained 6 DL auto-segmentation models based on 3-dimensional U-Net and residual 3-dimensional U-Net architectures. We then implemented an average and an optimally weighted average ensemble to improve prediction performance. We evaluated algorithms with the average and standard deviation of the volumetric Dice similarity coefficient, surface Dice similarity coefficient with 2- and 3-mm thresholds, and average symmetric surface distance. One independent expert radiation oncologist assessed the clinical viability of the DL contours and determined the necessary amount of editing before they could be used in clinical practice. RESULTS: Quantitatively, the ensembles performed the best across all structures. The optimal ensemble (volumetric Dice similarity coefficient, average symmetric surface distance) was (85.5 ± 6.4, 2.6 ± 0.8; GTV), (94.4 ± 1.5, 1.0 ± 0.4; gluteus maximus), (92.6 ± 0.9, 0.9 ± 0.1; gluteus medius), (85.0 ± 2.7, 1.1 ± 0.3; gluteus minimus), (92.1 ± 1.5, 0.8 ± 0.2; paraspinal), and (78.3 ± 5.7, 1.5 ± 0.6; piriformis). The qualitative evaluation suggested that the best model could reduce the total muscle and tumor delineation time to a 19-minute average. CONCLUSIONS: Our methodology produces expert-level muscle and sacral chordoma tumor segmentation using DL and ensemble modeling. It can substantially augment the streamlining and accuracy of treatment planning and represents a critical step toward automated delineation of the clinical target volume in sarcoma and other disease sites.

Congenital Heart Surgery Machine Learning-Derived In-Depth Benchmarking Tool.

BACKGROUND: We previously showed that machine learning-based methodologies of optimal classification trees (OCTs) can accurately predict risk after congenital heart surgery and assess case-mix-adjusted performance after benchmark procedures. We extend this methodology to provide interpretable, easily accessible, and actionable hospital performance analysis across all procedures. METHODS: The European Congenital Heart Surgeons Association Congenital Cardiac Database data subset of 172,888 congenital cardiac surgical procedures performed in European centers between 1989 and 2022 was analyzed. OCT models (decision trees) were built predicting hospital mortality (area under the curve [AUC], 0.866), prolonged postoperative mechanical ventilatory support time (AUC, 0.851), or hospital length of stay (AUC, 0.818), thereby establishing case-adjusted benchmarking standards reflecting the overall performance of all participating hospitals, designated as the "virtual hospital." OCT analysis of virtual hospital aggregate data yielded predicted expected outcomes (both aggregate and for risk-matched patient cohorts) for the individual hospital's own specific case-mix, readily available on-line. RESULTS: Raw average rates were hospital mortality, 4.9%; mechanical ventilatory support time, 14.5%; and length of stay, 15.0%. Of 146 participating centers, compared with each hospital's overall case-adjusted predicted hospital mortality benchmark, 20.5% statistically (<90% CI) overperformed and 20.5% underperformed. An interactive tool based on the OCT analysis automatically reveals 14 hospital-specific patient cohorts, simultaneously assessing overperformance or underperformance, and enabling further analysis of cohort strata in any chosen time frame. CONCLUSIONS: Machine learning-based OCT benchmarking analysis provides automatic assessment of hospital-specific case-adjusted performance after congenital heart surgery, not only overall but importantly, also by similar risk patient cohorts. This is a tool for hospital self-assessment, particularly facilitated by the user-accessible online-platform.

Development of a Machine Learning-Based Prescriptive Tool to Address Racial Disparities in Access to Care After Penetrating Trauma.

Importance: The use of artificial intelligence (AI) in clinical medicine risks perpetuating existing bias in care, such as disparities in access to postinjury rehabilitation services. Objective: To leverage a novel, interpretable AI-based technology to uncover racial disparities in access to postinjury rehabilitation care and create an AI-based prescriptive tool to address these disparities. Design, Setting, and Participants: This cohort study used data from the 2010-2016 American College of Surgeons Trauma Quality Improvement Program database for Black and White patients with a penetrating mechanism of injury. An interpretable AI methodology called optimal classification trees (OCTs) was applied in an 80:20 derivation/validation split to predict discharge disposition (home vs postacute care [PAC]). The interpretable nature of OCTs allowed for examination of the AI logic to identify racial disparities. A prescriptive mixed-integer optimization model using age, injury, and gender data was allowed to "fairness-flip" the recommended discharge destination for a subset of patients while minimizing the ratio of imbalance between Black and White patients. Three OCTs were developed to predict discharge disposition: the first 2 trees used unadjusted data (one without and one with the race variable), and the third tree used fairness-adjusted data. Main Outcomes and Measures: Disparities and the discriminative performance (C statistic) were compared among fairness-adjusted and unadjusted OCTs. Results: A total of 52 468 patients were included; the median (IQR) age was 29 (22-40) years, 46 189 patients (88.0%) were male, 31 470 (60.0%) were Black, and 20 998 (40.0%) were White. A total of 3800 Black patients (12.1%) were discharged to PAC, compared with 4504 White patients (21.5%; P < .001). Examining the AI logic uncovered significant disparities in PAC discharge destination access, with race playing the second most important role. The prescriptive fairness adjustment recommended flipping the discharge destination of 4.5% of the patients, with the performance of the adjusted model increasing from a C statistic of 0.79 to 0.87. After fairness adjustment, disparities disappeared, and a similar percentage of Black and White patients (15.8% vs 15.8%; P = .87) had a recommended discharge to PAC. Conclusions and Relevance: In this study, we developed an accurate, machine learning-based, fairness-adjusted model that can identify barriers to discharge to postacute care. Instead of accidentally encoding bias, interpretable AI methodologies are powerful tools to diagnose and remedy system-related bias in care, such as disparities in access to postinjury rehabilitation care.

Artificial intelligence versus surgeon gestalt in predicting risk of emergency general surgery.

BACKGROUND: Artificial intelligence (AI) risk prediction algorithms such as the smartphone-available Predictive OpTimal Trees in Emergency Surgery Risk (POTTER) for emergency general surgery (EGS) are superior to traditional risk calculators because they account for complex nonlinear interactions between variables, but how they compare to surgeons' gestalt remains unknown. Herein, we sought to: (1) compare POTTER to surgeons' surgical risk estimation and (2) assess how POTTER influences surgeons' risk estimation. STUDY DESIGN: A total of 150 patients who underwent EGS at a large quaternary care center between May 2018 and May 2019 were prospectively followed up for 30-day postoperative outcomes (mortality, septic shock, ventilator dependence, bleeding requiring transfusion, pneumonia), and clinical cases were systematically created representing their initial presentation. POTTER's outcome predictions for each case were also recorded. Thirty acute care surgeons with diverse practice settings and levels of experience were then randomized into two groups: 15 surgeons (SURG) were asked to predict the outcomes without access to POTTER's predictions while the remaining 15 (SURG-POTTER) were asked to predict the same outcomes after interacting with POTTER. Comparing to actual patient outcomes, the area under the curve (AUC) methodology was used to assess the predictive performance of (1) POTTER versus SURG, and (2) SURG versus SURG-POTTER. RESULTS: POTTER outperformed SURG in predicting all outcomes (mortality-AUC: 0.880 vs. 0.841; ventilator dependence-AUC: 0.928 vs. 0.833; bleeding-AUC: 0.832 vs. 0.735; pneumonia-AUC: 0.837 vs. 0.753) except septic shock (AUC: 0.816 vs. 0.820). SURG-POTTER outperformed SURG in predicting mortality (AUC: 0.870 vs. 0.841), bleeding (AUC: 0.811 vs. 0.735), pneumonia (AUC: 0.803 vs. 0.753) but not septic shock (AUC: 0.712 vs. 0.820) or ventilator dependence (AUC: 0.834 vs. 0.833). CONCLUSION: The AI risk calculator POTTER outperformed surgeons' gestalt in predicting the postoperative mortality and outcomes of EGS patients, and when used, improved the individual surgeons' risk prediction. Artificial intelligence algorithms, such as POTTER, could prove useful as a bedside adjunct to surgeons when preoperatively counseling patients. LEVEL OF EVIDENCE: Prognostic and Epidemiological; Level II.