Promoting breast health awareness: Can a sensor-enabled training system for patient education help?

INTRODUCTION: According to a 2009 study published in the Journal of Clinical Oncology, 79% of women (N = 222) diagnosed with breast cancer reported that they identified their cancers through breast self-exam (BSE). However, the U.S. Preventative Services Task Force does not require clinicians to teach women how to perform BSE. METHODS: To address this grave challenge, our team at the Technology Enabled Clinical Improvement (TECI) Center has developed a mobile, sensor-enabled haptic training system to teach women proper BSE technique. To validate the efficacy of the training system, our team deployed a data collection at the 2019 Breast Cancer and African Americans (BCAA) event where survey, sensor, and anecdotal data were collected from 61 participants. The custom-built breast model used in this study had a single, hard mass embedded in it. RESULTS: Participants at the BCAA event were able to successfully identify the mass 65% of the time and used an average force of 7.2 N. When looking at participants' confidence in their abilities to perform BSE, only 10% of respondents answered "very confident" pre-training whereas post-training, the reporting for "very confident" jumped to 66% (p < 0.01). CONCLUSION: By comparison, our previous work revealed that practitioners who use less than 10 N of force are 70% more likely to miss a lesion. The integration of sensors into the BSE haptic training system allowed for objective, evidence-based assessment of hands-on skill. In addition to teaching women proper BSE technique, this training empowered women to be informed advocates in their breast health journey. Future community-based training/feedback sessions will allow for continuous advancement of the training system.

Generating Rare Surgical Events Using CycleGAN: Addressing Lack of Data for Artificial Intelligence Event Recognition.

INTRODUCTION: Artificial Intelligence (AI) has shown promise in facilitating surgical video review through automatic recognition of surgical activities/events. There are few public video data sources that demonstrate critical yet rare events which are insufficient to train AI for reliable video event recognition. We suggest that a generative AI algorithm can create artificial massive bleeding images for minimally invasive lobectomy that can be used to augment the current lack of data in this field. MATERIALS AND METHODS: A generative adversarial network (GAN) algorithm was used (CycleGAN) to generate artificial massive bleeding event images. To train CycleGAN, six videos of minimally invasive lobectomies were utilized from which 1819 frames of nonbleeding instances and 3178 frames of massive bleeding instances were used. RESULTS: The performance of the CycleGAN algorithm was tested on a new video that was not used during the training process. The trained CycleGAN was able to alter the laparoscopic lobectomy images according to their corresponding massive bleeding images, where the contents of the original images were preserved (e.g., location of tools in the scene) and the style of each image is changed to massive bleeding (i.e., blood automatically added to appropriate locations on the images). CONCLUSIONS: The result could suggest a promising approach to supplement the lack of data for the rare massive bleeding event that can occur during minimally invasive lobectomy. Future work could be dedicated to developing AI algorithms to identify surgical strategies and actions that potentially lead to massive bleeding and warn surgeons prior to this event occurrence.

AI-Based Video Segmentation: Procedural Steps or Basic Maneuvers?

INTRODUCTION: Video-based review of surgical procedures has proven to be useful in training by enabling efficiency in the qualitative assessment of surgical skill and intraoperative decision-making. Current video segmentation protocols focus largely on procedural steps. Although some operations are more complex than others, many of the steps in any given procedure involve an intricate choreography of basic maneuvers such as suturing, knot tying, and cutting. The use of these maneuvers at certain procedural steps can convey information that aids in the assessment of the complexity of the procedure, surgical preference, and skill. Our study aims to develop and evaluate an algorithm to identify these maneuvers. METHODS: A standard deep learning architecture was used to differentiate between suture throws, knot ties, and suture cutting on a data set comprised of videos from practicing clinicians (N = 52) who participated in a simulated enterotomy repair. Perception of the added value to traditional artificial intelligence segmentation was explored by qualitatively examining the utility of identifying maneuvers in a subset of steps for an open colon resection. RESULTS: An accuracy of 84% was reached in differentiating maneuvers. The precision in detecting the basic maneuvers was 87.9%, 60%, and 90.9% for suture throws, knot ties, and suture cutting, respectively. The qualitative concept mapping confirmed realistic scenarios that could benefit from basic maneuver identification. CONCLUSIONS: Basic maneuvers can indicate error management activity or safety measures and allow for the assessment of skill. Our deep learning algorithm identified basic maneuvers with reasonable accuracy. Such models can aid in artificial intelligence-assisted video review by providing additional information that can complement traditional video segmentation protocols.

Do Individual Surgeon Preferences Affect Procedural Outcomes?

OBJECTIVES: Surgeon preferences such as instrument and suture selection and idiosyncratic approaches to individual procedure steps have been largely viewed as minor differences in the surgical workflow. We hypothesized that idiosyncratic approaches could be quantified and shown to have measurable effects on procedural outcomes. METHODS: At the American College of Surgeons (ACS) Clinical Congress, experienced surgeons volunteered to wear motion tracking sensors and be videotaped while evaluating a loop of porcine intestines to identify and repair 2 preconfigured, standardized enterotomies. Video annotation was used to identify individual surgeon preferences and motion data was used to quantify surgical actions. χ 2 analysis was used to determine whether surgical preferences were associated with procedure outcomes (bowel leak). RESULTS: Surgeons' (N=255) preferences were categorized into 4 technical decisions. Three out of the 4 technical decisions (repaired injuries together, double-layer closure, corner-stitches vs no corner-stitches) played a significant role in outcomes, P <0.05. Running versus interrupted did not affect outcomes. Motion analysis revealed significant differences in average operative times (leak: 6.67 min vs no leak: 8.88 min, P =0.0004) and work effort (leak-path length=36.86 cm vs no leak-path length=49.99 cm, P =0.001). Surgeons who took the riskiest path but did not leak had better bimanual dexterity (leak=0.21/1.0 vs no leak=0.33/1.0, P =0.047) and placed more sutures during the repair (leak=4.69 sutures vs no leak=6.09 sutures, P =0.03). CONCLUSIONS: Our results show that individual preferences affect technical decisions and play a significant role in procedural outcomes. Future analysis in more complex procedures may make major contributions to our understanding of contributors to procedure outcomes.

Diversity, equity, and inclusion in presidential leadership of academic medical and surgical societies.

BACKGROUND: Our aim was to identify gender and racial disparities in presidential leadership for national medical and surgical organizations. METHODS: We located publicly sourced information on national medical organizations. Years between or since the first diverse presidents were analyzed using descriptive statistics and Mann Whitney U tests. RESULTS: Sixty-seven national medical and surgical organizations were surveyed. 70.8% (n = 34) diversified via gender first (White-female), whereas 26.1% (n = 14) had racial diversity first. Organizations with gender diversity first followed with an African American male president sooner than organizations who first diversified by race (14.7 ± 11.8 v. 27.6 ± 11.3 years, p = 0.018). No significant difference was observed for the third tier of diversification. CONCLUSIONS: Significant gender and racial leadership disparities in national medical organizations are still present. It is notable that organizations with female leaders had a shorter timeline to racial diversity. These findings help to inform strategies to promote and increase diversity, equity, and inclusion in national leadership.

Developing a longitudinal database of surgical skills performance for practicing surgeons: A formal feasibility and acceptance inquiry.

BACKGROUND: We explored the feasibility and surgeons' perceptions of the utility of a longitudinal skills performance database. METHODS: A 10-station surgical skills assessment center was established at a national scientific meeting. Skills assessment volunteers (n = 189) completed a survey including opinions on practicing surgeons' skills evaluation, ethics, and interest in a longitudinal database. A subset (n = 23) participated in a survey-related interview. RESULTS: Nearly all participants reported interest in a longitudinal database and most believed there is an ethical obligation for such assessments to protect the public. Several interviewees specified a critical role for both formal and informal evaluation is to first create a safe and supportive environment. CONCLUSIONS: Participants support the construction of longitudinal skills databases that allow information sharing and establishment of professional standards. In a constructive environment, structured peer feedback was deemed acceptable to enhance and diversify surgeon skills. Large scale skills testing is feasible and scientific meetings may be the ideal location.

Can Deep Learning Algorithms Help Identify Surgical Workflow and Techniques?

BACKGROUND: Surgical videos are now being used for performance review and educational purposes; however, broad use is still limited due to time constraints. To make video review more efficient, we implemented Artificial Intelligence (AI) algorithms to detect surgical workflow and technical approaches. METHODS: Participants (N = 200) performed a simulated open bowel repair. The operation included two major phases: (1) Injury Identification and (2) Suture Repair. Accordingly, a phase detection algorithm (MobileNetV2+GRU) was implemented to automatically detect the two phases using video data. In addition, participants were noted to use three different technical approaches when running the bowel: (1) use of both hands, (2) use of one hand and one tool, or (3) use of two tools. To discern the three technical approaches, an object detection (YOLOv3) algorithm was implemented to recognize objects that were commonly used during the Injury Identification phase (hands versus tools). RESULTS: The phase detection algorithm achieved high precision (recall) when segmenting the two phases: Injury Identification (86 ± 9% [81 ± 12%]) and Suture Repair (81 ± 6% [81 ± 16%]). When evaluating three technical approaches in running the bowel, the object detection algorithm achieved high average precisions (Hands [99.32%] and Tools [94.47%]). The three technical approaches showed no difference in execution time (Kruskal-Wallis Test: P= 0.062) or injury identification (not missing an injury) (Chi-squared: P= 0.998). CONCLUSIONS: The AI algorithms showed high precision when segmenting surgical workflow and identifying technical approaches. Automation of these techniques for surgical video databases has great potential to facilitate efficient performance review.

Performance assessment using sensor technology.

Over the past 30 years, there have been numerous, noteworthy successes in the development, validation, and implementation of clinical skills assessments. Despite this progress, the medical profession has barely scratched the surface towards developing assessments that capture the true complexity of hands-on skills in procedural medicine. This paper highlights the development implementation and new discoveries in performance metrics when using sensor technology to assess cognitive and technical aspects of hands-on skills.

Multimodal Cardiopulmonary Bypass Skills Assessment Within a High-Fidelity Simulation Environment.

BACKGROUND: A high-fidelity simulator that uses a perfused porcine heart, cannulae, and tubing has been demonstrated to be a useful training adjunct. We hypothesized that multimodal assessment of cardiopulmonary bypass (CPB) skills within this high-fidelity simulated environment could discern expert from trainee performance. METHODS: Three traditional fellows (postgraduate year 6-8) and 3 attending surgeons each performed 3 aortic cannulations. The third sequence included venous cannulation, commencement of CPB, and placement of a cardioplegia catheter and aortic cross-clamp. Performance across 20 cognitive and 21 technical domains was evaluated. Surgeon and assistant hand movements and economy of motion were assessed by electromagnetic motion sensors worn under sterile gloves. RESULTS: Analysis showed a significant difference in cognitive (6.7 ± 2.3 vs 4.6 ± 2.7, P = .03) but not technical (6.2 ± 2.5 vs 5.8 ± 2.2, P = .7) scores favoring the experts. In addition, experts showed higher efficiency by spending 64 ± 14 seconds to construct a nonpledgeted aortic purse-string suture and secure it with a Rummel, while trainees spent 82 ± 30 seconds to complete this task (P = .03). Motion analysis revealed similar path lengths between experts and trainees for cannulation and CPB but significantly shorter path lengths for experts in cross-clamp (47.5 ± 15.5 m vs 91.9 ± 20.3 m, P = .04). CONCLUSIONS: Multimodal assessment using cognitive, technical, and motion analysis of basic CPB tasks using a high-fidelity simulation environment is a valid system to measure performance and discriminate experts from trainees. This construct may allow for development of "competence thresholds" with important implications for training and certification in cardiothoracic surgery.

Translating motion tracking data into resident feedback: An opportunity for streamlined video coaching.

BACKGROUND: We hypothesized that differences in motion data during a simulated laparoscopic ventral hernia repair (LVH) can be used to stratify top and lower tier performers and streamline video review. MATERIALS AND METHODS: Surgical residents (N = 94) performed a simulated partial LVH repair while wearing motion tracking sensors. We identified the top ten and lower ten performers based on a final product quality score (FPQS) of the repair. Two blinded raters independently reviewed motion plots to identify patterns and stratify top and lower tier performers. RESULTS: Top performers had significantly higher FPQS (23.3 ± 1.2 vs 5.7 ± 1.6 p < 0.01). Raters identified patterns and stratified top performers from lower tier performers (Rater 1 χ2 = 3.2 p = 0.07 and Rater 2 χ2 = 2.0 p = 0.16). During video review, we correlated motion plots with the relevant portion of the procedure. CONCLUSION: Differences in motion data can identify learning needs and enable rapid review of surgical videos for coaching.

Use of sensors to quantify procedural idle time: Validity evidence for a new mastery metric.

BACKGROUND: Quantification of mastery is the first step in using objective metrics for teaching. We hypothesized that during orotracheal intubation, top tier performers have less idle time compared to lower tier performers. METHODS: At the Anesthesiology 2018 Annual Meeting, 82 participants intubated a normal airway simulator and a burnt airway simulator. The movements of the participant's laryngoscope were quantified using electromagnetic motion sensors. Top tier performers were defined as participants who intubated both simulators successfully in less than the median time for each simulator. Idle time was defined as the duration of time when the laryngoscope was not moving. RESULTS: Top performers showed less Idle Time when intubating the normal airway compared to lower tier performers (14.5 ± 9.8 seconds vs 34.0 ± 52.0 seconds, respectively P < .01). Likewise, top performers showed less Idle Time when intubating the burnt airway compared to lower tier performers (18.6 ± 15.2 seconds vs 63.4 ± 59.11 seconds; P < .01). Comparing performance on the burnt airway to the normal airway, there was a difference for lower tier performers (63.4 ± 59.1 seconds vs 34.0 ± 52.0 seconds; P < .01) but not for top tier performers (18.6 ± 15.2 seconds vs 14.5 ± 9.8 seconds; P = .07). CONCLUSION: Similar to our previous findings with other procedures, Idle Time was shown to have known group validity evidence when comparing top performers with lower tier performers. Further, Idle Time was correlated with procedure difficulty in our prior work. We observed statistically significant differences in Idle Times for lower tier performers when comparing the normal airway to the burnt airway but not for top tier performers. Our findings support the continued exploration of Idle Time for development of objective assessment and curricula.

Screening surgical residents' laparoscopic skills using virtual reality tasks: Who needs more time in the sim lab?

BACKGROUND: This study investigated the possibility of using virtual reality perceptual-motor tasks as a screening tool for laparoscopic ability. We hypothesized that perceptual-motor skills assessed using virtual reality will correlate with the quality of simulated laparoscopic ventral hernia repair. MATERIALS AND METHODS: Surgical residents (N = 37), performed 2 virtual reality perceptual-motor tasks: (1) force matching and (2) target tracking. Participants also performed a laparoscopic ventral hernia repair on a simulator and final product quality score, and endoscopic visualization errors were calculated. Correlational analysis was performed to assess the relationship between performance on virtual reality tasks and laparoscopic ventral hernia repair. RESULTS: Residents with poor performance on force matching in virtual reality-"peak deflection" (r = -0.34, P < .05) and "summation distance" (r = -0.36, P < .05)-had lower final product quality scores. Likewise, poor performance in virtual reality-based target tracking-"path length" (r = -0.49, P < .05) and "maximum distance" (r = -0.37, P < .05)-correlated with a lower final product quality score. CONCLUSION: Our findings support the notion that virtual reality could be used as a screening tool for perceptual-motor skill. Trainees identified as having poor perceptual-motor skill can benefit from focused curricula, allowing them to hone personal areas of weakness and maximize technical skill.

Residents' surgical performance during the laboratory years: an analysis of rule-based errors.

BACKGROUND: Nearly one-third of surgical residents will enter into academic development during their surgical residency by dedicating time to a research fellowship for 1-3 y. Major interest lies in understanding how laboratory residents' surgical skills are affected by minimal clinical exposure during academic development. A widely held concern is that the time away from clinical exposure results in surgical skills decay. This study examines the impact of the academic development years on residents' operative performance. We hypothesize that the use of repeated, annual assessments may result in learning even without individual feedback on participants simulated performance. METHODS: Surgical performance data were collected from laboratory residents (postgraduate years 2-5) during the summers of 2014, 2015, and 2016. Residents had 15 min to complete a shortened, simulated laparoscopic ventral hernia repair procedure. Final hernia repair skins from all participants were scored using a previously validated checklist. An analysis of variance test compared the mean performance scores of repeat participants to those of first time participants. RESULTS: Twenty-seven (37% female) laboratory residents provided 2-year assessment data over the 3-year span of the study. Second time performance revealed improvement from a mean score of 14 (standard error = 1.0) in the first year to 17.2 (SD = 0.9) in the second year, (F[1, 52] = 5.6, P = 0.022). Detailed analysis demonstrated improvement in performance for 3 grading criteria that were considered to be rule-based errors. There was no improvement in operative strategy errors. CONCLUSIONS: Analysis of longitudinal performance of laboratory residents shows higher scores for repeat participants in the category of rule-based errors. These findings suggest that laboratory residents can learn from rule-based mistakes when provided with annual performance-based assessments. This benefit was not seen with operative strategy errors and has important implications for using assessments not only for performance analysis but also as a learning experience.