Surgeon Performance Research Articles

Intraoperative AI-assisted early prediction of parathyroid and ischemia alert in endoscopic thyroid surgery.

The preservation of parathyroid glands is crucial in endoscopic thyroid surgery to prevent hypocalcemia and related complications. However, current methods for identifying and protecting these glands have limitations. We propose a novel technique that has the potential to improve the safety and efficacy of endoscopic thyroid surgery. Our study aims to develop a deep learning model called PTAIR 2.0 (Parathyroid gland Artificial Intelligence Recognition) to enhance parathyroid gland recognition during endoscopic thyroidectomy. We compare its performance against traditional surgeon-based identification methods. Parathyroid tissues were annotated in 32 428 images extracted from 838 endoscopic thyroidectomy videos, forming the internal training cohort. An external validation cohort comprised 54 full-length videos. Six candidate algorithms were evaluated to select the optimal one. We assessed the model's performance in terms of initial recognition time, identification duration, and recognition rate and compared it with the performance of surgeons. Utilizing the YOLOX algorithm, we developed PTAIR 2.0, which demonstrated superior performance with an AP50 score of 92.1%. The YOLOX algorithm achieved a frame rate of 25.14 Hz, meeting real-time requirements. In the internal training cohort, PTAIR 2.0 achieved AP50 values of 94.1%, 98.9%, and 92.1% for parathyroid gland early prediction, identification, and ischemia alert, respectively. Additionally, in the external validation cohort, PTAIR outperformed both junior and senior surgeons in identifying and tracking parathyroid glands (p < 0.001). The AI-driven PTAIR 2.0 model significantly outperforms both senior and junior surgeons in parathyroid gland identification and ischemia alert during endoscopic thyroid surgery, offering potential for enhanced surgical precision and patient outcomes.

Belonging in Surgery: A Validated Instrument and Single Institutional Pilot.

The aim of this study was to develop and validate an instrument to measure Belonging in Surgery among surgical residents. Belonging is the essential human need to maintain meaningful relationships and connections to one's community. Increased belongingness is associated with better well-being, job performance, and motivation to learn. However, no tools exist to measure belonging among surgical trainees. A panel of experts adapted a belonging instrument for use among United States surgery residents. After administration of the 28-item instrument to residents at a single institution, a Cronbach alpha was calculated to measure internal consistency, and exploratory principal component analyses were performed. Multiple iterations of analyses with successively smaller item samples suggested the instrument could be shortened. The expert panel was reconvened to shorten the instrument. Descriptive statistics measured demographic factors associated with Belonging in Surgery. The overall response rate was 52% (114 responses). The Cronbach alpha among the 28 items was 0.94 (95% CI: 0.93-0.96). The exploratory principal component analyses and subsequent Promax rotation yielded 1 dominant component with an eigenvalue of 12.84 (70% of the variance). The expert panel narrowed the final instrument to 11 items with an overall Cronbach alpha of 0.90 (95% CI: 0.86, 0.92). Belonging in Surgery was significantly associated with race (Black and Asian residents scoring lower than White residents), graduating with one's original intern cohort (residents who graduated with their original class scoring higher than those that did not), and inversely correlated with resident stress level. An instrument to measure Belonging in Surgery was validated among surgical residents. With this instrument, Belonging in Surgery becomes a construct that may be used to investigate surgeon performance and well-being.

Algorithm Versus Expert: Machine Learning Versus Surgeon-Predicted Symptom Improvement After Carpal Tunnel Release.

Surgeons rely on clinical experience when making predictions about treatment effects. Incorporating algorithm-based predictions of symptom improvement after carpal tunnel release (CTR) could support medical decision-making. However, these algorithm-based predictions need to outperform predictions made by surgeons to add value. We compared predictions of a validated prediction model for symptom improvement after CTR with predictions made by surgeons. This cohort study included 97 patients scheduled for CTR. Preoperatively, surgeons estimated each patient's probability of improvement 6 months after surgery, defined as reaching the minimally clinically important difference on the Boston Carpal Tunnel Syndrome Symptom Severity Score. We assessed model and surgeon performance using calibration (calibration belts), discrimination (area under the curve [AUC]), sensitivity, and specificity. In addition, we assessed the net benefit of decision-making based on the prediction model's estimates vs the surgeon's judgement. The surgeon predictions had poor calibration and suboptimal discrimination (AUC 0.62, 95%-CI 0.49-0.74), while the prediction model showed good calibration and appropriate discrimination (AUC 0.77, 95%-CI 0.66-0.89, P = .05). The accuracy of surgeon predictions was 0.65 (95%-CI 0.37-0.78) vs 0.78 (95%-CI 0.67-0.89) for the prediction model (P = .03). The sensitivity of surgeon predictions and the prediction model was 0.72 (95%-CI 0.15-0.96) and 0.85 (95%-CI 0.62-0.97), respectively (P = .04). The specificity of the surgeon predictions was similar to the model's specificity (P = .25). The net benefit analysis showed better decision-making based on the prediction model compared with the surgeons' decision-making (ie, more correctly predicted improvements and/or fewer incorrectly predicted improvements). The prediction model outperformed surgeon predictions of improvement after CTR in terms of calibration, accuracy, and sensitivity. Furthermore, the net benefit analysis indicated that using the prediction model instead of relying solely on surgeon decision-making increases the number of patients who will improve after CTR, without increasing the number of unnecessary surgeries.

Publication performance of university maxillofacial surgeons in Germany

The aim of this bibliometric analysis was to benchmark the publication activities of German university departments of oral and maxillofacial surgery.The publication performance of staff surgeons (chief and consultants), documented by first or last authorship, from 37 German university departments was captured over a 10-year period (January 1, 2010, to December 31, 2019). All publications listed in PubMed were included. Additionally, the Impact Factor (IF) was determined.A total of 213 surgeons were identified, of whom 158 (74.2%) were publishing. The number of publications was 1,777, published in 311 journals. Publication activity ranged from an average of 23.3 publications per staff surgeon in the top-ranked department to 0 publications in the last-ranked. The same trend was observed for the total cumulative IFs (CIFs) per member (range from 56.2 to 0). The most common used journal was the Journal of Cranio-Maxillofacial Surgery (19.7%), with focus on “dentoalveolar surgery” (24%) and “operative techniques and procedures” (28.3%). Women constituted 19.2% of the staff, contributing to 8.5% of the publications.The publication performance of German university departments of oral and maxillofacial surgery exhibits a high variance, which did not correlate with the number of personnel and could only be explained by different research motivations.

Review of "What Happens on Call Doesn't Stay on Call. The Effects of In-House Call on Acute Care Surgeons' Sleep and Burnout: Results of the Surgeon Performance Trial" by Coleman JJ et al. Annals of Surgery 2023;278:497-505.

Review of "What Happens on Call Doesn't Stay on Call. The Effects of In-House Call on Acute Care Surgeons' Sleep and Burnout: Results of the Surgeon Performance Trial" by Coleman JJ et al. Annals of Surgery 2023;278:497-505.

Evidence-Based Performance Measures for Reconstruction after Skin Cancer Resection: A Multidisciplinary Performance Measure Set.

The American Society of Plastic Surgeons commissioned the multidisciplinary Performance Measure Development Work Group on Reconstruction after Skin Cancer Resection to identify and draft quality measures for the care of patients undergoing skin cancer reconstruction. Included stakeholders were the American Academy of Otolaryngology-Head and Neck Surgery, the American Academy of Facial Plastic and Reconstructive Surgery, the American Academy of Dermatology, the American Society of Dermatologic Surgery, the American College of Mohs Surgery, the American Society for Mohs Surgery, and a patient representative. Two outcome measures and five process measures were identified. The outcome measures included the following: (1) patient satisfaction with information provided by their surgeon before their facial procedure, and (2) postprocedural urgent care or emergency room use. The process measures focus on antibiotic stewardship, anticoagulation continuation and/or coordination of care, opioid avoidance, and verification of clear margins. All measures in this report were approved by the American Society of Plastic Surgeons Quality and Performance Measures Work Group and Executive Committee, and the stakeholder societies. The work group recommends the use of these measures for quality initiatives, Continuing Medical Education, Continuous Certification, Qualified Clinical Data Registry reporting, and national quality reporting programs.

Practical Guide to Surgical Simulation Research.

This Guide to Statistics and Methods provides an overview of simulation-based education and its impact on surgeon performance.

Two different methods to digitally visualize continuous electrocochleography potentials during cochlear implantation: a first description of feasibility

PurposeThe study explores the potential of real-time electrocochleographic potentials (ECochG) visualization during electrode insertion using digital microscopes such as RoboticScope (BHS®). Collaborative software development of the MAESTRO Software (MED-EL®) offers continuous ECochG monitoring during implantation and postoperative hearing evaluation, addressing previous time constraints. The study aims to assess software applicability and the impact of real-time visualization on long-term residual hearing preservation.MethodsEight patients with residual hearing underwent cochlear implantation with Flex26 or Flex28 electrode according to the Otoplan evaluation. ECochG responses were measured and visualized during electrode insertion, with insertion times recorded. Two randomized display methods (graph and arrows) tracked ECochG potentials. Postoperative behavioral thresholds determined hearing preservation. Successful real-time intraoperative ECochG visualization was achieved in all cases, enabling surgeon adaptation. Mean electrode insertion time was 114 s, with postoperative thresholds comparable to preoperative values. Visualization did not affect surgeon workload. ECochG amplitudes differed between patients with and without residual hearing.ConclusionThe study demonstrates effective implementation of advanced ECochG software combined with real-time visualization, enabling residual hearing preservation during CI. Visualization had no apparent effect on surgeon performance or workload. Future investigation involving a larger population will assess the long-term impact of ECochG on hearing threshold and structure preservation.

Performance outcomes from a DMEK peeling and preparation wet lab

ObjectiveTo evaluate the Descemet membrane endothelial keratoplasty (DMEK) preparation performance of trainee surgeons in an ex vivo human donor cornea DMEK wet lab simulation setting.MethodsHuman donor corneoscleral rims unsuitable for...

Examining early career surgeon performance: the most common procedures and their associated complication rates

Examining early career surgeon performance: the most common procedures and their associated complication rates

The role of irrigation fluid in transurethral resection of the prostate outcomes and surgeon performance.

Transurethral resection of the prostate (TURP) is the gold-standard for the treatment of benign prostate enlargement (BPE) associated with lower urinary tract symptoms (LUTS), after failure of conservative therapy. At present, only resection-rate (grams of prostate resected over time) is regarded as an efficiency parameter to evaluate the skill of the operator and to assess the outcome of the procedure. Five surgeons performed TURP using a Gyrus-type bipolar system in 123 patients with BPE/LUTS who came to our observation from June 2016 to December 2019. The amount of irrigation fluid used during the procedure was registered and correlated to the operating time, resection-rate, prostate adenoma weight, post-operative bladder irrigation time, intraoperative bleeding and days of catheterization. We found an inverse correlation between the amount of irrigation fluid used during TURP and the resection-rate recorded for all operators, according to Spearman's Correlation (r = -0.78, p = 0.002); a direct correlation was also found between the amount of irrigation fluid and the adenoma weight. Finally, we also found a direct correlation with intraoperative bleeding and the duration and amount of bladder irrigation during and after TURP. The amount of irrigation fluid used is proposed as a reliable parameter to estimate the efficiency of the endoscopic procedure as well to assess the skill of the operator and shortterm results. The observed data encourage the possibility of applying this new efficiency indicator to all endoscopic maneuvers.

A comparative cadaveric study for percutaneous scaphoid fixation: robotic vs freehand.

To compare the robotic-assisted and the traditional freehand percutaneous scaphoid fixation in number of guidewire attempts, duration of fluoroscopy time, amount of radiation dose, and screw centrality. Twenty cadaveric specimens were randomized into either the robotic or freehand group. The scaphoids in both groups were fixed by either the same attending or resident from our hand surgery department. The operation duration, amount of radiation from intraoperative fluoroscopy, total fluoroscopy time, and the number of guidewire attempts were documented and compared. Postoperatively, all the specimens had a computed tomography (CT) scan performed, and the difference in the final position of the screw and the central axis of the scaphoid was examined. In the robotic group, all the guide wires were satisfactorily positioned within a single attempt, while the median number of attempts in the traditional freehand group was 18 (quaternion 14-65). This also meant that the surgeon in the robotic group experienced significantly lower radiation exposure dose and time as compared to the freehand group. There were no significant differences in the final screw position as compared to the central axis of the scaphoid in both groups. Although there was no difference in surgeon performance in the robotic group, the operative time for the attending was significantly lower as compared to the resident in the freehand group. Robotic-assisted surgery for scaphoid fracture fixation is superior to the traditional freehand method as it facilitates accurate screw placement with lower radiation exposure and fewer guide wire attempts.

The Influence of Patient and Surgeon Characteristics on Press Ganey Provider Ratings in Plastic Surgery.

Although patient satisfaction ratings are increasingly used as hospital and provider performance metrics, these ratings may be affected by factors extraneous to surgeon performance. The aim of this study was to assess whether outpatient Press Ganey ratings for plastic surgery providers were tied more to provider or patient characteristics. All Consumer Assessment of Healthcare Providers and Systems Clinician & Group Survey (CG-CAHPS) responses for plastic surgery providers from 2017 to 2023 from a single institution were analyzed. Ordered logistic regression models were used to identify characteristics associated with provider ratings. In total, 6442 surveys from 4594 patients representing 29 plastic surgery providers were analyzed. Across the patient cohort, 23.2% (N = 1492) rated plastic surgery providers lower than a 10/10. After adjustment, provider characteristics including provider gender, patient-provider gender concordance, and years in practice were not associated with overall provider ratings (all P > .05). In contrast, patient characteristics were associated with provider ratings. Older patient age (odds ratio [OR], 1.02; P < .001) was associated with higher ratings. Finally, self-reported overall health and mental health were associated with lower provider ratings (both P < .05); specifically, mental health categories lower than "Excellent" were associated with significantly lower provider ratings (overall mental health "Very Good': OR, 0.57; "Good": OR, 0.53; "Fair": OR, 0.59; "Poor": OR, 0.55; all P < .05). Press Ganey ratings for plastic surgeons are influenced by patient characteristics, including age and mental health. Because satisfaction ratings are increasingly used as performance metrics in plastic surgery, adjusted rating models may be needed to more accurately reflect surgeon performance.

Efficiency of a polymer-coated phacoemulsification tip in cataract surgery.

To determine whether there is a difference in efficiency during phacoemulsification when using the hybrid polymer tip compared with the metal balanced tip, as measured by cumulative dissipated energy (CDE) and total procedure time. John A. Moran Eye Center, University of Utah, Salt Lake City, Utah. Experimental study. Phacoemulsification was performed with the Centurion Vision System and Infiniti OZil handpiece, with a 30-degree bevel metal balanced tip and hybrid polymer tip. The vacuum was set at 500 mm Hg and the aspiration flow rate at 50 mL/min. Power was changed for torsional and longitudinal ultrasound (US) at 20%, 40%, 60%, 80%, and 100%. Hardened porcine lenses were cut into 200 two mm cubes, 10 for each specified tip/power setting. They were placed into an artificial anterior chamber filled with a balanced salt solution, and phacoemulsification was performed. A second observer randomly assigned the US type and power. The same observer timed trials with a stopwatch, and CDE was recorded. There were no significant differences in time or CDE for any power levels or US modes when comparing the 2 tips. No significant difference in efficiency was observed when comparing the metal balanced tip with the hybrid polymer tip. Given the known improved safety profile and the overall equivalent efficiency of the hybrid tip, it could improve the performance of beginning and advanced surgeons in the future. Future study is needed to evaluate other setting combinations and to evaluate the hybrid tip in clinical settings.

Assessment of robotic telesurgery system among surgeons: a single-center study

The field of robotic-assisted surgery is expanding rapidly; therefore, future robotic surgeons will need to be trained in an organized manner. Here, we aimed to examine surgeon performance on the Sinaflex Robotic Telesurgery System for correlation with training hours spent in training program. This is a prospective study of a single-center experience at the Hasan Sadikin Hospital, Bandung City of West Java, Indonesia. We included 43 surgeons from 11 departments, all invited to train using the Sinaflex Robotic Telesurgery system at the Hasan Sadikin Hospital. All study cohorts have never performed a robotic surgery procedure beforehand and have had at least five years of field experience. The surgeons were free to choose their training duration and simulation. After finishing the training session, they were asked to perform several tasks with increasing difficulty levels. There were nine training tasks in total with increasing levels of difficulty. A total of 43 surgeons from 11 different department were included in this prospective study. Our study was separated into 3 different batches and most surgeons failed to pass the examination (n = 12, 8, and 9, for batches 1, 2, and 3, respectively). The “failed” surgeon, additionally, tended to be older than the “passed” cohort (49.3 ± 7.4 vs 42.1 ± 7.3 years old, p = 0.005). In terms of duration of hours spent training on the robot, there was little difference training hours between the cohort that passed and the cohort that failed cohort (10.0 [8.4–10.1] vs 10.0 [8.0–10.0], respectively) with a p value of 0.265. We found no correlation between the total hours spent in the training program and surgeon performance on the Sinaflex robotic telesurgery system. Structured robot surgical training courses must be incorporated into the training programs.

Performance of Orthopaedic Shoulder and Elbow Surgeons on a Biostatistical Knowledge Examination.

The objective of this study is to evaluate the biostatistical interpretation abilities of fellowship trained orthopaedic surgeons. A cross-sectional survey was administered to orthopaedic surgeon members of the American Shoulder and Elbow Surgeons (ASES), assessing orthopaedic surgeon attitudes towards biostatistics, confidence in understanding biostatistics, and ability to interpret biostatistical measures on a multiple-choice test. A 4.5% response rate was achieved with 55 complete survey responses. The mean percent correct was 55.2%. Higher knowledge test scores were associated with younger age and fewer years since board exam completion (p ≤ 0.001). Greater average number of publications per year correlated with superior statistical interpretation (p=0.009). Respondents with higher self-reported confidence were more likely to accurately interpret results (p ≤ 0.017). Of the respondents, 93% reported frequently using statistics to form medical opinions, 98% answered that statistical competency is important in the practice of orthopaedic surgery, and 80% were eager to continue learning biostatistics. It is concerning that fellowship-trained shoulder and elbow surgeons, many of whom frequently publish or are reviewing scientific literature for publication, are scoring 55.2% correctly on average on this biostatistical knowledge examination. Surgeons that are further from formal statistical knowledge training are more likely to have lower biostatistical knowledge test scores. Respondents who published at the highest rate were associated with higher scores. Continuing medical education in biostatistics may be beneficial for maintaining statistical knowledge utilised in the current literature.

Force Profile as Surgeon-Specific Signature.

To investigate the notion that a surgeon's force profile can be the signature of their identity and performance. Surgeon performance in the operating room is an understudied topic. The advent of deep learning methods paired with a sensorized surgical device presents an opportunity to incorporate quantitative insight into surgical performance and processes. Using a device called the SmartForceps System and through automated analytics, we have previously reported surgeon force profile, surgical skill, and task classification. However, an investigation of whether an individual surgeon can be identified by surgical technique has yet to be studied. In this study, we investigate multiple neural network architectures to identify the surgeon associated with their time-series tool-tissue forces using bipolar forceps data. The surgeon associated with each 10-second window of force data was labeled, and the data were randomly split into 80% for model training and validation (10% validation) and 20% for testing. Data imbalance was mitigated through subsampling from more populated classes with a random size adjustment based on 0.1% of sample counts in the respective class. An exploratory analysis of force segments was performed to investigate underlying patterns differentiating individual surgical techniques. In a dataset of 2819 ten-second time segments from 89 neurosurgical cases, the best-performing model achieved a micro-average area under the curve of 0.97, a testing F1-score of 0.82, a sensitivity of 82%, and a precision of 82%. This model was a time-series ResNet model to extract features from the time-series data followed by a linearized output into the XGBoost algorithm. Furthermore, we found that convolutional neural networks outperformed long short-term memory networks in performance and speed. Using a weighted average approach, an ensemble model was able to identify an expert surgeon with 83.8% accuracy using a validation dataset. Our results demonstrate that each surgeon has a unique force profile amenable to identification using deep learning methods. We anticipate our models will enable a quantitative framework to provide bespoke feedback to surgeons and to track their skill progression longitudinally. Furthermore, the ability to recognize individual surgeons introduces the mechanism of correlating outcome to surgeon performance.

Pilot Analysis of Surgeon Instrument Utilization Signatures Based on Shannon Entropy and Deep Learning for Surgeon Performance Assessment in a Cadaveric Carotid Artery Injury Control Simulation.

Assessment and feedback are critical to surgical education, but direct observational feedback by experts is rarely provided because of time constraints and is typically only qualitative. Automated, video-based, quantitative feedback on surgical performance could address this gap, improving surgical training. The authors aim to demonstrate the ability of Shannon entropy (ShEn), an information theory metric that quantifies series diversity, to predict surgical performance using instrument detections generated through deep learning. Annotated images from a publicly available video data set of surgeons managing endoscopic endonasal carotid artery lacerations in a perfused cadaveric simulator were collected. A deep learning model was implemented to detect surgical instruments across video frames. ShEn score for the instrument sequence was calculated from each surgical trial. Logistic regression using ShEn was used to predict hemorrhage control success. ShEn scores and instrument usage patterns differed between successful and unsuccessful trials (ShEn: 0.452 vs 0.370, P < .001). Unsuccessful hemorrhage control trials displayed lower entropy and less varied instrument use patterns. By contrast, successful trials demonstrated higher entropy with more diverse instrument usage and consistent progression in instrument utilization. A logistic regression model using ShEn scores (78% accuracy and 97% average precision) was at least as accurate as surgeons' attending/resident status and years of experience for predicting trial success and had similar accuracy as expert human observers. ShEn score offers a summative signal about surgeon performance and predicted success at controlling carotid hemorrhage in a simulated cadaveric setting. Future efforts to generalize ShEn to additional surgical scenarios can further validate this metric.

Development of a prospective biliary dashboard to compare performance and surgical cost.

Transparency around surgeon level data may align healthcare delivery with quality care for patients. Biliary surgery includes numerous procedures performed by both general surgeons and subspecialists alike. Cholecystectomy is a common surgical procedure and an optimal cohort to measure quality outcomes within a healthcare system. Data were collected for 5084 biliary operations performed by 68 surgeons in 11 surgical divisions in a health system including a tertiary academic hospital, two regional community hospitals, and two ambulatory surgery centers. A privacy protected dashboard was developed to compare surgeon performance and cost between July 2018 and June 2022. A sample cohort of patients ≥ 18years who underwent cholecystectomy were compared by operative time, cost, and 30-day outcomes. Over 4years, 4568 cholecystectomy procedures were performed by 57 surgeons. Operations were done by 57 surgeons in four divisions and included 3846 (84.2%) laparoscopic cholecystectomies, 601 (13.2%) laparoscopic cholecystectomies with cholangiogram, and 121 (2.6%) open cholecystectomies. Patients were admitted from the emergency room in 2179 (47.7%) cases while 2389 (52.3%) cases were performed in the ambulatory setting. Individual surgeons were compared to peers for volume, intraoperative data, cost, and outcomes. Cost was lowest at ambulatory surgery centers, yet only 4.2% of elective procedures were performed at these facilities. Prepackaged kits with indocyanine green were more expensive than cholangiograms that used iodinated contrast. The rate of emergency department visits was lowest when cases were performed at ambulatory surgery centers. Data generated from clinical dashboards can inform surgeons as to how they compare to peers regarding quality metrics such as cost, time, and complications. In turn, this may guide strategies to standardize care, optimize efficiency, provide cost savings, and improve outcomes for cholecystectomy procedures. Future application of clinical dashboards can assist surgeons and administrators to define value-based care.

SP11.8 Deep learning for performance assessment of laparoscopic sleeve gastrectomy (LSG): validation against objective structured assessment of technical skills (OSATS) and clinical outcomes

Abstract Background Current methods of surgeon performance assessment are unstructured, labour-intensive and subjective. Machine learning could provide rapid, automated and subjective assessment. We design a custom deep learning model to output performance metrics from videos of LSG and validate these against OSATS and clinical outcomes. Methods A novel dataset of 3210 images from videos of LSG was annotated to train an instrument segmentation framework based on Mask R-CNN, a state-of-the-art deep learning algorithm. Spatial outputs of detected instruments were used to determine workflow through a Markov chain and secondly to determine instrument metrics for assessment of surgical performance. Metrics including time spent in stage and instrument time, distance, smoothness, concentration, and efficiency were then validated against OSATS ratings from 2 independent scorers and clinical outcomes from 35 LSGs. Results Grasper and ligasure efficiency and smoothness were higher in videos rated within the top quartile by OSATS score when compared to those in the bottom quartile and when comparing consultants to trainees. Videos in the top quartile for grasper smoothness had higher total OSATS scores (17.31±2.27) compared to those in the lowest quartile (15.69±2.93) and across all 5 individual OSATS domains in addition to higher percentage of total weight loss at 3, 6, and 12 months. Conclusions This study demonstrates how vision based deep learning can be used for surgeon performance assessment. Future work will aim to validate this against a large number of videos, expand upon the range of metrics produced, and look towards the steps needed for clinical translation.