Objectives Although the long-term prognosis after lung transplantation has improved recently, primary graft dysfunction (PGD) remains the major cause of early mortality. The aim of this study was to elucidate trends in PGD incidence and short-term mortality following lung transplantation in the contemporary era. Methods We analyzed a single-center database of lung transplantations performed across three periods (Era 1: 2009–2013, Era 2: 2014–2017, and Era 3: 2018–2021). PGD was graded according to the 2016 International Society for Heart and Lung Transplantation definition, and PGD grade 3 within T0–T72 was used as the primary outcome. Trends in PGD incidence, factors associated with PGD, and early mortality rates after lung transplantation were identified. Results This study included 856 lung transplants: 277 in Era 1, 296 in Era 2, and 283 in Era 3. PGD grade 3 incidence decreased significantly over time: 35.9% (99 cases) in Era 1, 26.4% (78 cases) in Era 2, and 18.4% (52 cases) in Era 3 (P<0.001). During the study period, the lung allocation score (LAS) and intraoperative cardiopulmonary bypass (CPB) use decreased, whereas the use of intraoperative nitric oxide and extracorporeal membrane oxygenation increased. Logistic multivariate modeling identified era, recipient sex (male), underlying disease, race, and blood transfusion as factors associated with PGD. No significant difference was observed in 30-day hospital mortality across the three eras (2.9%, 1.4%, and 1.4% for Era 1, Era 2, and Era 3, respectively; P=0.313). Conclusion This study demonstrated a significant reduction in PGD incidence over time, which coincided with a decrease in LAS and intraoperative CPB use. However, no significant changes were observed in short-term mortality after lung transplantation.

BACKGROUND:Thoracic organ procurement procedures have been standardized for decades, and serious organ injury during procurement is considered rare. However, when injuries do occur, they may result in irreversible loss of transplantable organs. CASE SUMMARY:We report a rare case of unintentional donor heart injury caused by transdiaphragmatic liver core needle biopsy using a Tru-Cut-type device during multi-organ procurement. Two puncture injuries were identified on the inferior wall of the heart adjacent to the posterior descending coronary artery. Although there was no active bleeding and gross ventricular function appeared preserved, the proximity to a coronary branch raised concern for potential coronary injury and re-bleeding following systemic heparinization. After multidisciplinary discussion among the cardiac procurement team, recipient transplant team, and the organ procurement organization, the donor heart was declined intraoperatively. CONCLUSION:This case highlights a preventable mechanism of procurement-related cardiac injury and underscores the importance of coordination, timing, and situational awareness among procurement teams to minimize avoidable donor organ loss.

OBJECTIVE:Lung transplantation is the definitive treatment for end-stage pulmonary disease, but ongoing challenges remain in long-term survival. We report our single-center experience of 2000 adult lung transplants over a nearly 35-year period and assess trends in patient demographics, intraoperative management, and perioperative and long-term outcomes. METHODS:We retrospectively reviewed 2000 lung transplants performed between 1988 and 2023 at our center. Recipients and donors were separated into 3 eras: Era 1 (1988-2000), Era 2 (2001-2011), and Era 3 (2012-2023). Recipient outcomes were compared among the eras. RESULTS:There were differences in recipient demographics across the eras. Over time, we have increasingly performed transplantation in patients with restrictive lung disease. Overall graft survival has improved, with median graft survival increasing from 5.5 years (Era 1) to 9.0 years (Era 3) (P < .0001). We observed similar trends when patients were stratified by transplant indication. The incidence of primary graft dysfunction grade 3 has remained stable at 28.7% in Era 2 and 26.7% in Era 3 (P = .4892). The median freedom from chronic lung allograft dysfunction has improved from 3.2 years (Era 2) to 3.4 years (Era 3) (P = .0001). CONCLUSIONS:Lung graft survival has improved over time at our institution due to advances in perioperative and long-term management. However, primary graft dysfunction grade 3 rates have not changed, and chronic lung allograft dysfunction is commonly diagnosed within the first 4 years after transplant. Further research is necessary to understand these disease processes and to generate new treatment strategies to address them.

OBJECTIVE:The use of lungs from brain-dead donors is low partly as a result of the lack of reliable donor assessment criteria. The validated Lung Donor (LUNDON) score predicts lung acceptance for transplantation by using 9 clinically relevant variables, including the presence of an abnormality on radiograph of the chest. Because most organ donor evaluations now include routine computed tomography (CT) of the chest, we aimed to assess whether the addition of CT findings impacts the LUNDON model's performance. METHODS:Data including CT findings were collected for adult brain-dead donors from 3 organ procurement organizations from 2014 to 2020. The primary outcome was lung acceptance for transplantation. We collated all CT findings into a weighted CT composite score, with greater scores representing more CT abnormalities, and calculated the score for each donor. RESULTS:The lung acceptance rate was 40.4% among 2454 donors with CTs of the chest and 22.3% among 1980 donors without CTs of the chest. Emphysema, pulmonary edema, and traumatic lung injury on CT were associated with a lower likelihood lung acceptance. The LUNDON model's performance was comparable between use of the original radiograph of the chest variable, the CT composite score, or both variables together (C-statistics 0.883, 0.887, 0.890, respectively). All 3 iterations of the model were predictive of 1-year graft survival. CONCLUSIONS:Undergoing CT was independently associated with donor lung acceptance. The incorporation of highly granular findings from CT of the chest to the previously established LUNDON model maintained, but did not meaningfully improve, its excellent baseline ability to predict lung use and its association with graft survival.

Background The dire consequences of heart failure (HF) patient non-response to guideline directed medical therapy often fuel early, non-selective referral for surgical intervention (ventricular assist device [VAD] or transplant). The high-risk associated with these interventions mandates precision in directing them only toward those patients who would otherwise suffer severe near-term deterioration. We previously reported a 52,265-patient deep learning model that predicted 1-year severe decompensation/death in HF inpatients, with a C-statistic of 0.91. We now present external model validation. Few groups applying deep learning to large-scale datasets have achieved external validation using equally large-scale independent datasets, yet proof of generalization is essential to practical applicability. Methods Our previous study used standard electronic health record (EHR) data to build ensemble deep learning models employing time-series and densely connected networks. The positive-class included both all-cause mortality and referral for HF surgical intervention within 1 year. In the current study, we assessed generalization of model architecture in an external validation test set from the Veterans Cardiac Health and Artificial Intelligence Model Predictions (V-CHAMPS) challenge, a synthetic national governmental sample using a distinct EHR system. While V-CHAMPS is a robust dataset, variables that capture VAD/transplant referral were not readily extracted, limiting the positive-class to mortality only. Results A total of 380,441 distinct admissions from 75,086 HF patients contributed >720 million EHR datapoints. 23% of observations fit positive-class criteria. The model C-statistic in the external-validation cohort was 0.79. Conclusions Despite being developed in a single-center dataset with a more precise positive-class, our model architecture maintained relative accuracy when applied to a national sample in an unrelated EHR system. This supports clinical relevancy of the deep-learning model and adaptability with retraining to disparate contexts. This broad applicability suggests considerable potential of EHR-based deep learning models to assist HF clinicians in improving the usage of advanced surgical therapy. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement No external funding was received for this study. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: This study was approved by the Washington University School of Medicine Human Studies Institutional Review Board and was performed using a synthetic data lake from the US Department of Veterans Affairs. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes VHA Synthetic Data Lake (V-CHAMPS)

Background Early identification of heart failure patients at increased risk for near-term adverse outcomes would assist clinicians in efficient resource allocation and improved care. Deep learning can improve identification of these patients. Methods This retrospective study examined adult heart failure patients admitted to a tertiary care institution between January 2009 and December 2018. A deep learning model was constructed with a dense input layer, three long short-term memory (LSTM) layers, and a dense hidden layer to cohesively extract features from time-series and non-time-series EHR data. Primary outcomes were all-cause hospital readmission or death within 30 days after hospital discharge. Results Among a final subset of 49,675 heart failure patients, we identified 171,563 hospital admissions described by 330 million EHR data points. There were 22,111 (13%) admissions followed by adverse 30-day outcomes, including 19,122 readmissions (87%) and mortality in 3,330 patients (15%). Our final deep learning model achieved an area under the receiver-operator characteristic curve (AUC) of 0.613 and precision-recall (PR) AUC of 0.38. Conclusions This EHR-based deep learning model developed from a decade of heart failure care achieved marginal clinical accuracy in predicting very early hospital readmission or death despite previous accurate prediction of 1-year mortality in this large study cohort. These findings suggest that factors unavailable in standard EHR data play pivotal roles in influencing early hospital readmission. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement No external funding received. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Not Applicable The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The human studies Institutional Review Board at Washington University School of Medicine approved this study. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Not Applicable I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Not Applicable I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Not Applicable Data is available upon request. Provision of patient-protected data is not possible to protect patient identity.

BACKGROUND Primary graft dysfunction (PGD) is the leading cause of death in the first 30 days after lung trans-plantation and is also associated with worse long-term outcomes. Outcomes of patients with PGD grade 3 requiring extracorporeal membrane oxygenation (ECMO) support after lung transplantation have yet to be well described. We sought to describe short-and long-term outcomes for patients with PGD grade 3 who required ECMO support.METHODS This is a single-center retrospective cohort study of patients undergoing lung transplantation. We stratified patients with PGD grade 3 into non-ECMO, venoarterial (VA) ECMO, and venovenous (VV) ECMO groups after trans-plantation. We then compared the outcomes between the groups.RESULTS Of 773 lung transplant recipients, PGD grade 3 developed in 204 (26%) at any time in the first 72 hours after lung transplantation. Of these, 13 (5%) required VA ECMO and 25 (10%) required VV ECMO support. The 30-day, 1-year, and 5-year survival in the VA ECMO group was 62%, 54%, and 43% compared with 96%, 84%, and 65% in the VV ECMO group and 99%, 94%, and 71% in the non-ECMO group. Multivariable Cox regression analysis showed that VA ECMO was associated with increased mortality (hazard ratio, 2.37; 95% CI, 1.06-5.28; P [ .04).CONCLUSIONS Patients who required VA ECMO support for PGD grade 3 have significantly worse survival compared with those who did not require ECMO and those who required VV ECMO support. This suggests that VA ECMO treatment of patients with PGD grade 3 after lung transplantation can be a predictable risk factor for mortality.(Ann Thorac Surg 2023;115:1273-81)(c) 2023 by The Society of Thoracic Surgeons

Background. Pulmonary carcinoid tumorlet (PCT) is defined as small proliferation of neuroendocrine cells that invade the adjacent basement membrane. It is often associated with chronic pulmonary inflammatory processes. However, the characteristics of PCT in end-stage lung diseases remain unclear. Methods. We conducted a retrospective cohort study of the explanted lungs after transplantation at our institution between January 1999 and October 2020. Patients who underwent re-transplantation were excluded. Results. Pulmonary carcinoid tumorlet was incidentally discovered in the explanted lungs from 15 patients (1.1%) out of 1367 lung transplants performed during the study period. Nine patients (60.0 %) were women, with a median age of 59 years (IQR: 57-62) at transplant. Underlying pulmonary indications for lung transplantation were chronic obstructive pulmonary disease (9/15, 60.0%), interstitial lung disease (2/15, 13.0%), pulmonary vascular disease (2/15, 13.0%), alpha-1 antitrypsin deficiency (1/15, 7.0%), and bronchiectasis (1/15, 7.0%). Of the patients who underwent bilateral lung transplantation (13/15, 86.7%), PCT was found in the right lung in 10 patients (10/13, 76.9%). Thirteen patients had one lesion, 1 patient had 2 lesions and 1 patient had multiple lesions. Conclusion. Our study shows that PCT is generally uncommon, but when it occurs, it occurs more frequently on the right side and in female patients with end-stage pulmonary disease. Chronic obstructive pulmonary disease may be a predisposing factor for developing PCT.

There is a chronic shortage of donor lungs for pulmonary transplantation due, in part, to low lung utilization rates in the United States. We performed a retrospective cohort study using data from the Scientific Registry of Transplant Recipients database (2006-2019) and developed the lung donor (LUNDON) acceptability score. A total of 83 219 brain-dead donors were included and were randomly divided into derivation (n = 58 314, 70%) and validation (n = 24 905, 30%) cohorts. The overall lung acceptance was 27.3% (n = 22 767). Donor factors associated with the lung acceptance were age, maximum creatinine, ratio of arterial partial pressure of oxygen to fraction of inspired oxygen, mechanism of death by asphyxiation or drowning, history of cigarette use (≥20 pack-years), history of myocardial infarction, chest x-ray appearance, bloodstream infection, and the occurrence of cardiac arrest after brain death. The prediction model had high discriminatory power (C statistic, 0.891; 95% confidence interval, 0.886-0.895) in the validation cohort. We developed a web-based, user-friendly tool (available at https://sites.wustl.edu/lundon) that provides the predicted probability of donor lung acceptance. LUNDON score was also associated with recipient survival in patients with high lung allocation scores. In conclusion, the multivariable LUNDON score uses readily available donor characteristics to reliably predict lung acceptability. Widespread adoption of this model may standardize lung donor evaluation and improve lung utilization rates.

OBJECTIVE:National and institutional data suggest an increase in organ discard rate (donor lungs procured but not implanted) after a new lung allocation policy was introduced in 2017. However, this measure does not include on-site decline rate (donor lungs declined intraoperatively). The objective of this study is to examine the impact of the allocation policy change on on-site decline. METHODS:We used a Washington University (WU) and our local organ procurement organization (Mid-America Transplant [MTS]) database to abstract data on all accepted lung offers from 2014 to 2021. An on-site decline was defined as an event in which the procuring team declined the organs intraoperatively, and the lungs were not procured. Logistic regression models were used to investigate potentially modifiable reasons for decline. RESULTS:The overall study cohort comprised 876 accepted lung offers, of which 471 donors were at MTS with WU or others as the accepting center and 405 at other organ procurement organizations with WU as the accepting center. At MTS, the on-site decline rate increased from 4.6% to 10.8% (P = .01) after the policy change. Given the greater likelihood of non-local organ placement and longer travel distance after policy change, the estimated cost of each on-site decline increased from $5727 to $9700. In the overall group, latest partial pressure of oxygen (odds ratio [OR], 0.993; 95% confidence interval [CI], 0.989-0.997), chest trauma (OR, 2.474; CI, 1.018-6.010), chest radiograph abnormality (OR, 2.902; CI, 1.289-6.532), and bronchoscopy abnormality (OR, 3.654; CI, 1.813-7.365) were associated with on-site decline, although lung allocation policy era was unassociated (P = .22). CONCLUSIONS:We found that nearly 8% of accepted lungs are declined on site. Several donor factors were associated with on-site decline, although lung allocation policy change did not have a consistent impact on on-site decline.

The Accreditation Council for Graduate Medical Education works diligently to ensure that educational metric thresholds are consistently met in all cardiothoracic (CT) surgical training programs. You can be assured that an outstanding educational experience is available in yours. So, why do some trainees exit with so much more than others? The good news is that much is left in your hands. Although all trainees start out with the best intentions, it is behavior, not intention, that determines outcome (Figure 1).

Background Studies in lung transplantation have shown variable association between hospital volume and clinical outcomes. We aimed to identify the pattern of effect of hospital volume on individual patient survival after lung transplantation. Methods We performed a retrospective analysis using the United Network for Organ Sharing national thoracic organ transplantation database. Adult patients who underwent lung transplantation between January 2013 and December 2017 were included. The association between mean annual center volume and 1-year overall survival was examined using restricted cubic splines in a random effects multivariable Cox model. The volume threshold for optimal 1-year overall survival was subsequently approximated by the maximum likelihood approach using segmented linear splines in the same model. Results The study included 10,007 patients at 71 transplant centers. Median annual center volume was 22 cases (interquartile range, 10.6 to 38). A center volume threshold was identified at 33 cases per year (95% confidence interval, 28 to 37). Higher center volume, to 33 cases per year, was associated with better 1-year survival (hazard ratio 0.989, 95% confidence interval, 0.980 to 0.999 every additional case). Further increase in center volume above 33 cases per year showed no additional benefit (hazard ratio 1.000, 95% confidence interval, 0.996 to 1.003 every additional case). Twenty-three centers (32.4%) reached the volume threshold of 33 cases per year. Conclusions One-year survival after lung transplantation improved with increasing center volume to as many as 33 cases per year. Low volume centers below the 33 cases per year threshold had large variations in their outcomes and had a higher risk of performing poorly, although many of them maintained good performance.

Partial anomalous pulmonary venous return is a rare congenital aberrancy that involves oxygen-rich pulmonary venous drainage into the right atrium instead of into the systemic circulation. This report describes a case of isolated partial anomalous pulmonary venous return of the right upper lobe in a donor lung. Successful transplantation was performed with a Carrel patch technique for left atrial cuff reconstruction using a segment of donor vena cava. This report of partial anomalous pulmonary venous return in a right donor lung describes this reconstructive approach to restore physiologic venous drainage.

BACKGROUND Surgical mechanical ventricular assistance and cardiac replacement therapies, although life-saving in many heart failure (HF) patients, remain high-risk. Despite this, the difficulty in timely identification of medical therapy nonresponders and the dire consequences of nonresponse have fueled early, less selective surgical referral. Patients who would have ultimately responded to medical therapy are therefore subjected to the risk and life disruption of surgical therapy. OBJECTIVES The purpose of this study was to develop deep learning models based upon commonly-available electronic health record (EHR) variables to assist clinicians in the timely and accurate identification of HF medical therapy nonresponders. METHODS The study cohort consisted of all patients (age 18 to 90 years) admitted to a single tertiary care institution from January 2009 through December 2018, with International Classification of Disease HF diagnostic coding. Ensemble deep learning models employing time-series and densely-connected networks were developed from standard EHR data. The positive class included all observations resulting in severe progression (death from any cause or referral for HF surgical intervention) within 1 year. RESULTS A total of 79,850 distinct admissions from 52,265 HF patients met observation criteria and contributed > 350 million EHR datapoints for model training, validation, and testing. A total of 20% of model observations fit positive class criteria. The model C-statistic was 0.91. CONCLUSIONS The demonstrated accuracy of EHR-based deep learning model prediction of 1-year all-cause death or referral for HF surgical therapy supports clinical relevance. EHR-based deep learning models have considerable potential to assist HF clinicians in improving the application of advanced HF surgical therapy in medical therapy nonresponders. (c) 2022 the American College of Cardiology Foundation. Published by Elsevier. All rights reserved.

BACKGROUND Continuous-flow left ventricular assist device (CF-LVAD) support is a mainstay in the hemodynamic management of patients with end-stage heart failure refractory to optimal medical therapy. In this report we evaluated waitlist complications and competing outcomes for CF-LVAD patients compared with primary transplant candidates listed for orthotopic heart transplantation at a single center.METHODS All patients listed for orthotopic heart transplantation between 2006 and 2020 at our institution were retrospec-tively reviewed (CF-LVAD, 300; primary transplant, 244). Kaplan-Meier methodology with log-rank testing was used to evaluate survival outcomes. Terminal outcomes of death, delisting, and transplant were assessed as competing risks and compared between groups using Gray's test. Multivariable Fine-Gray regression was used to identify predictors of transplantation.RESULTS One-year rates of transplant, delisting, and death were 48%, 8%, and 2%, respectively, for CF-LVAD patients and 45%, 15%, and 9%, respectively, for primary transplant (all P < .001). Waitlist mortality at 5 years was 4% among CF-LVAD patients and 13% for primary transplants. All-cause mortality after listing was lower for CF-LVAD patients (P = .017). There was no difference in posttransplant survival between groups (P = .250). On multivariable Fine-Gray regression stroke (P = .017), respiratory failure (P = .032), right ventricular failure (P = .019), and driveline infection (P = .050) were associated with decreased probability of transplantation. Posttransplant survival was not significantly worse for CF-LVAD patients who experienced device-related complications (P = .901).CONCLUSIONS Although device-related complications were significantly associated with decreased rates of trans-plant, CF-LVAD patients had excellent waitlist outcomes overall. In light of the 2018 allocation score change the risk of complications should be taken into account when deciding whether to offer CF-LVAD as a bridge to transplant.(Ann Thorac Surg 2022;114:1276-83)(c) 2022 by The Society of Thoracic Surgeons

Objective: The decision to perform single lung transplants or double lung transplants is usually made before the operation. We have previously reported that a proportion of single lung transplants were unexpectedly performed in the setting of an aborted double lung transplant, and these patients may be at a higher risk of worse short-term outcomes. Long-term outcomes in unplanned single lung transplants remain unknown. Methods: We analyzed a single-center database of lung transplants from 2000 to 2020. Single lung transplants were classified into planned and unplanned groups after reviewing operative notes. Root cause analysis was performed for unplanned single lung transplants. Results: Of the 1326 lung transplants, 1265 (95%) were double lung transplants and 61 (5%) were single lung transplants (22 planned [36%], 39 unplanned [64%]). Underlying indications for transplant were significantly different; planned single lung transplant: chronic obstructive pulmonary disease (55%) and idiopathic pulmonary fibrosis (45%); unplanned single lung transplants: chronic obstructive pulmonary disease (23%), idiopathic pulmonary fibrosis (39%), and bronchiolitis obliterans syndrome (13%). The primary reasons for unplanned single lung transplant were donor-related (3, 7.7%), recipient-related (31, 80%), and donor and recipientrelated factors (5, 13%). Unplanned single lung transplants were more likely to require cardiopulmonary bypass during the operation (planned: 4/22, 18% vs unplanned: 20/39, 51%) but had shorter ischemic times (planned: 251 +/- 58 minutes vs unplanned: 221 +/- 48 minutes). The 5-year overall survival was 53% in the planned and 58% in the unplanned groups, respectively (P = .323). No difference in chronic lung allograft dysfunction-free survival (P = .995) was observed. Conclusions: Unplanned single lung transplants in the setting of aborted double lung transplant may be associated with acceptable long-term outcomes.

Background. Acute interstitial pneumonia (AIP), also known as Hamman-Rich syndrome, is a rare and rapidly progressive idiopathic interstitial lung disease with a high mortality rate. Treatment is limited to supportive care and empirical high-dose steroids; however, outcomes are generally poor. There are few reports of lung transplantation (LTx) in patients with AIP. Methods. We retrospectively identified patients with AIP among those who underwent LTx at our center between January 2008 and December 2020. Results. During the study period, 4 patients with AIP underwent bilateral LTx: 3 men and 1 woman, between 30 and 57 years of age. The lung allocation score ranged between 71 and 89. Of the 4 patients, 2 needed extracorporeal membrane oxygenation and mechanical ventilation (MV) and 1 needed MV preoperatively. Time of onset to transplant ranged from 1 to 3 months. None of the patients had primary graft dysfunction after LTx; 2 had acute cellular rejection and 1 had chronic lung allograft dysfunction. The 4 patients are alive with survival ranging between 1 and 12 years after LTx. Conclusion. AIP should be considered in patients with acute respiratory failure without a clear etiology. Our study showed that LTx led to good outcomes and should be considered as a treatment option in appropriate candidates.

Background: Machine learning models have potential to identify non-intuitive and previously unrecognized relationships between standardized clinical variables and the clinical manifestation of pathophysiological conditions. We used machine learning to examine the association of Society of Thoracic Surgeons (STS) Database variables with the presence of clinically significant ischemic mitral regurgitation (IMR) in patients undergoing coronary artery bypass grafting (CABG).Methods: STS Database variables (n=53) served as predictors of clinically significant IMR in machine learning modeling of 7,005 patients extracted from our institutional STS Database [1996–2018] who underwent CABG only (negative class, n=6,642) or CABG plus mitral valve intervention (positive class, n=363). Data were randomly partitioned into training (5,604 total patients, 281 positive, 5,323 negative) and test sets (1,401 total patients, 82 positive, 1,319 negative). The Synthetic Minority Oversampling Technique (SMOTE) was employed to produce a balanced training set.Results: Machine learning models, including random forests (RF), support vector machines (SVM), logistic regression (LR), and deep neural networks (DNN), were tested. Following training, final models predicted class labels for the patients in the test set. The models predicted class labels with promising accuracy (area under the receiver operating characteristic curve (AUC) values: RF, 0.70; SVM, 0.80; LR, 0.79; DNN, 0.80).Conclusions: STS Database variables have a predictive association with the presence of clinically significant IMR in patients undergoing surgical revascularization. These readily available variables may have potential as predictive variables in future translational machine learning modeling to assist in directing surgical care.