Generative Artificial Intelligence (GenAI) is increasingly being used in medical education, including the creation of content for clinical virtual patient simulation (VPS). We share practical insights drawn from our experience as medical educators, designed to help other educators integrate GenAI tools such as ChatGPT, Gemini, or Claude. We provide 12 practical tips that illustrate how GenAI can be used for the creation of content for VPS to streamline case development and increase realism for our learners. GenAI saves resources of time and expense by streamlining case creation, including creating patient images, clinical reports, and assessment questions. Clinical content experts can now concentrate on defining the case focus and on final-stage review, improving efficiency. It simplifies the development of effective scoring rubrics and meaningful formative feedback, enhancing learner engagement and clinical reasoning. By reducing the time needed to create VPS content, GenAI can allow educators to focus more on educational objectives and feedback to programs and learners. While GenAI is new and has limitations, these are likely to be addressed in the near future. As GenAI evolves, it will continue to transform VPS through improved realism, streamlined case creation, and more informative feedback, further supporting its role in medical education.

While enhanced breast screening of germline pathogenic variant (GPV) carriers results in earlier stage at diagnosis, the impact of tumour biology and GPV on chemotherapy receipt in early-stage disease remains understudied. We retrospectively reviewed treatment administered following a first diagnosis of BRCA1/2- and PALB2-associated breast cancer between 2002 and 2022. Chemotherapy receipt was compared according to tumor size, biologic subtype, and GPV. Subgroup analyses were performed in women with T1N0 disease and in those with pre-diagnostic awareness of their GPV. Overall, 309 affected BRCA1/2 and PALB2 carriers with a median age of 43 years at breast cancer diagnosis (range, 19–80 years) were included; 160 (51.8

Importance:Between 5% and 10% of breast cancer cases are associated with an inherited germline pathogenic or likely pathogenic variant (GPV) in a breast cancer susceptibility gene (BCSG), which could alter local and systemic therapy recommendations. Traditional genetic testing criteria misses a proportion of these cases. Objective:To evaluate the prevalence and clinicopathological associations of GPVs in 2 groups of BCSGs among an ethnically diverse cohort of women with newly diagnosed breast cancer. Design, Setting, and Participants:This cross-sectional study, conducted at 3 Montreal hospitals between September 2019 and April 2022, offered universal genetic counseling and testing to all women with a first diagnosis of invasive breast cancer. Women were offered an obligatory primary panel of BRCA1, BRCA2, and PALB2 (B1B2P2) and an optional secondary panel of 14 additional BCSGs. Eligible participants were women 18 years of age or older who received a diagnosis of a first primary invasive breast cancer not more than 6 months before the time of referral to the study. Data were analyzed from November 2023 to June 2024. Results:Of 1017 referred patients, 805 were eligible and offered genetic counseling and testing, and 729 of those 805 (90.6%) consented to be tested. The median age at breast cancer diagnosis was 53 years (range, 23-91 years), and 65.4% were White and of European ancestry. Fifty-four GPVs were identified in 53 patients (7.3%), including 39 patients (5.3%) with B1B2P2 and 15 patients (2.1%) with 6 of the 14 secondary panel BCSGs (ATM, BARD1, BRIP1, CHEK2, RAD51D, and STK11). On multivariable analysis, clinical factors independently associated with B1B2P2-positive status included being younger than 40 years of age at diagnosis (odds ratio [OR], 6.83; 95% CI, 2.22-20.90), triple-negative breast cancer (OR, 3.19; 95% CI, 1.20-8.43), high grade disease (OR, 1.68; 95% CI, 1.05-2.70), and family history of ovarian cancer (OR, 9.75; 95% CI, 2.65-35.85). Of 39 B1B2P2-positive patients, 13 (33.3%) were eligible for poly(adenosine diphosphate-ribose) polymerase (PARP) inhibitors. Conclusions and Relevance:In this cross-sectional universal genetic testing study of women with newly diagnosed invasive breast cancer, the prevalence of GPVs was 7.3%, with 5.3% of patients testing positive for B1B2P2. Among B1B2P2-women women, one-third were eligible for PARP inhibitors.

BACKGROUND AND PURPOSE:Preoperative partial breast irradiation (PBI) is a novel technique that can be used in patients with early-stage breast cancer with the goal of limiting the irradiated breast volume, toxicity and number of fractions. The aim of this trial is to assess the toxicity, surgical, oncologic and cosmetic outcomes of preoperative PBI. MATERIALS AND METHODS:In this single-arm phase II trial, we enrolled women ≥ 60 years, with unifocal low-risk breast invasive ductal carcinoma (cT1N0, grade 1-2, ER+, Her2-). Patients were treated with a single fraction of 20 Gy of preoperative PBI using volumetric modulated arc therapy (VMAT). Patients then underwent breast-conserving surgery (BCS) +/- sentinel lymph node biopsy within 72 h of radiation. Primary outcomes were rate of surgical complications and early toxicity. Secondary outcomes were cosmesis at 12 months, chronic toxicity and ipsilateral breast tumor recurrence. RESULTS:Twenty-five patients were recruited with a median age of 67 years, and a median follow-up of 60 months. Sentinel biopsy was positive in 1 out of 24 patients (4 %). Two patients received adjuvant RT for close margins or positive lymph nodes. Within the first 90 days, none of the patients had surgical complications; almost all had grade 0 to 1 acute and late RTOG skin toxicity. The cosmetic outcome was rated between good and excellent in all cases by physicians and patients, except for one patient who self-rated her cosmesis as fair as of the third year. There were no recurrences. CONCLUSION:Preoperative single-fraction PBI is a safe and feasible treatment for elderly patients with low-risk early-stage breast cancer, with no surgical complications, very low rates of acute and late radiation toxicity, and excellent cosmetic outcomes. Randomized controlled trials are needed to compare preoperative to adjuvant PBI in this patient population.

Abstract Introduction: While enhanced breast screening of germline BRCA1/2 carriers results in earlier stage at diagnosis, the impact of tumour biology and BRCA mutation on chemotherapy receipt in early stage disease remains understudied. Methods: We retrospectively reviewed treatment administered following a first diagnosis of BRCA1/2-associated breast cancer between 2003-2023 at our institution. Chemotherapy receipt (neoadjuvant or adjuvant) was evaluated according to tumor size, biologic subtype, and BRCA mutation. Current guidelines for PARP inhibitor use were applied to estimate the proportion of affected BRCA1/2 carriers that would be deemed eligible for targeted therapy in the future. Results: Overall, 251 affected BRCA1/2 carriers were included; 137 (54.6%) BRCA1 (median age 40 years, range 19-72) and 114 (45.4%) BRCA2 (median age 43, range, 24-80 years). Chemotherapy was administered in 70.1% of index breast cancer cases and was significantly associated with clinical tumor size (36.7% T1a-T1b, 90.9% T1c, 95.2% T2, 95.3% T3-T4, p< 0.001), nodal status (71.8% cN0 vs. 100% cN1-2, p=0.004), and biologic subtype (90.0% TNBC vs. 75.0% ER+HER2-, p=0.02). BRCA1-associated breast cancers were less likely to present with DCIS or T1 tumours (%Tis/T1; 46.7% BRCA1 vs. 70.8% BRCA2, p< 0.001) and more likely to present with triple negative disease (71.4% BRCA1 vs. 24.6% BRCA2, p< 0.001). BRCA1 carriers were more likely to require chemotherapy for index breast cancer (81.8% BRCA1 vs. 56.1% BRCA2, p< 0.001). In subgroup analysis of early stage, T1N0 disease, chemotherapy was administered in 79.0% BRCA1 and 52.2% BRCA2 patients (p=0.03). If recent guidelines incorporating biologic subtype, nodal involvement, and response to neoadjuvant chemotherapy were retrospectively applied to the cohort, 33.6% would be deemed eligible for PARP inhibitors in the adjuvant setting, including 40.9% BRCA1 and 17.5% BRCA2 affected carriers (p < 0.001). Conclusion: Chemotherapy receipt is high in BRCA-associated breast cancers including in early stage, node-negative disease. Overall, one third of affected carriers are expected to be eligible for PARP inhibitors in the adjuvant setting. Future studies exploring how this information may impact decisions around risk-reducing mastectomy are warranted. Citation Format: Stephanie Wong, Carla Apostolova, Amina Ferroum, Basmah Alhassan, Ipshita Prakash, Mark Basik, Karyne Martel, Sarkis Meterissian, David Fleiszer, Nora Wong, Talia Malagon, William Foulkes, Jean-Francois Boileau. Evaluating chemotherapy receipt and candidacy for PARP inhibitors in germline BRCA1/2 carriers with early and locally advanced breast cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-08-07.

Background:Virtual patient simulations are interactive, computer-based cases. We designed scenarios based on the McGill Simulation Complexity Score (MSCS), a previously described objective complexity score. We aimed to establish validity of the MSCS and introduce a novel learning tool in trauma education at our institution. Methods:After design of an easy and difficult patient scenario, we randomized medical students and residents to each perform 1 of the 2 scenarios. We conducted a 2-way analysis of variance of training level (medical student, resident) and scenario complexity (easy, difficult) to assess their effects on virtual time, the number of steps taken in the scenario, beneficial and harmful actions, and the ratio of beneficial over harmful actions. Results:Virtual patient scenarios were successfully designed using the MSCS. Twenty-four medical students and 12 residents participated in the easy scenario (MSCS = 3), and 27 medical students and 12 residents did the difficult scenario (MSCS = 18). Though beneficial actions were similar between students and residents, sudents performed more harmful actions, particularly when the scenario was difficult. One virtual patient died in the easy scenario and 3 died in the difficult one (all medical students). Performance varied with level of complexity and there was significant interaction between level of training and number of steps, as well as with number of harmful actions. Decreasing performance with increasing level of complexity, as defined by the MSCS, suggests this score can accurately quantify difficulty. Conclusion:We established validity of the MSCS and showed its successful application on virtual patient scenario design.

# 01. Near-peer tutoring: an effective adjunct for virtual anatomy learning {#article-title-2} The start of the COVID-19 pandemic caused a shift in medical education from the classroom to the virtual setting. This abrupt change led to an increase in stress among students. In response, McGill

Background: The Queen Bee phenomenon describes the behavioural response that occurs when women achieve success in a male-dominated environment, and in this position of authority, treat their female subordinates more critically. This behavioural response has been demonstrated in fields such as business, academia, the military, and police force. The goal of this study was to determine whether the Queen Bee phenomenon occurs similarly in surgical specialties within medicine. We hypothesized that female surgeons, fellows, and senior surgical residents would be more critical in their assessment of junior female residents than of their male counterparts. Methods: A scenario-based survey was distributed to all Canadian surgical programs, guised as a survey-based study of patterns of interaction among surgical trainees and staff surgeons. Scenario questions were designed to assess either female or male learners. Respondents’ average Likert score for female-based and male-based questions were calculated. The differences in scores were analyzed. Subgroup analyses were performed based on age, seniority, and surgical specialty. Results: We collected 716 survey responses, with 387 respondents identifying as male (54%) and 321 identifying as female (45%); the remaining 8 respondents did not report gender. This included 385 staff surgeons (54%), 66 fellows (9%), and 263 residents (37%); the remaining 2 respondents did not identify their group. The mean Likert scores for female respond ents assessing female learners was significantly lower than when male learners were assessed ( p = 0.008). During sub-group analysis, resident females scored female learners significantly less favourably than the staff and fellow groups ( p = 0.036). Some subspecialties demonstrated significant scoring differences. Female orthopedic surgeons scored female learners more harshly than male learners ( p = 0.049). Conclusion: The Queen Bee phenomenon was shown

# 01. The Queen Bee phenomenon in Canadian surgical subspecialties: an evaluation of gender biases in the resident training environment {#article-title-2} The Queen Bee phenomenon describes the behavioural response that occurs when women achieve success in a male-dominated environment, and in this

Purpose: Patients with early stage breast cancer (ESBC) are conventionally treated with breast-conserving surgery (BCS) followed by whole-breast external beam radiation therapy (EBRT). The emergence of targeted intraoperative radiation therapy (TARGIT) with Intrabeam has been used as a therapeutic alternative for patients with risk-adapted ESBC. Here we present our radiation therapy toxicities (RTT), postoperative complications (PC), and short-term outcomes of the prospective phase II trial at the McGill UniversityMethods and Materials: Patients aged =50 years with biopsy-proven hormone receptor-positive, grade 1 or 2, invasive ductal carcinoma of the breast, cT1N0, were eligible for the study. Enrolled patients underwent BCS followed by immediate TARGIT of 20 Gy in 1 fraction. Upon final pathology, patients with low-risk breast cancer (LRBC) received no further EBRT, and those with high-risk breast cancer (HRBC) received further 15 to 16 fractions of whole breast EBRT. HRBC criteria included pathologic tumor size >2 cm, grade 3, positive lympho-vascular invasion, multifocal disease, close margins (<2 mm), or positive nodal disease.Results: A total of 61 patients with ESBC were enrolled in the study; upon final pathology, 40 (65.6%) had LRBC, and 21 (34.4%) had HRBC. The median follow-up was 3.9 years. The most common HRBC criteria were close margins in 66.6% (n = 14) and lymphovascular invasion in 28.6% (n = 6). No grade 4 RTT were observed in either group. The most common PC were seroma and cellulitis for both groups. The rate of locoregional recurrence was 0% in both groups. The overall survival in LRBC was 97.5% and in HRBC 95.2% with no significant differences. Deaths were nonbreast cancer related.Conclusions: In patients with ESBC undergoing BCS, the use of TARGIT shows low rates of RTT and PC complications. Moreover, our short-term outcomes show no significant difference at 3.9 years median follow-up for locoregional recurrence or overall survival

Background: In the midst of the North American opioid crisis, identifying and intervening on drivers of high-risk opioid prescriptions is an important step towards reducing iatrogenic harm. Objectives: We aimed to identify factors associated with variations in high-risk opioid discharge prescriptions, following select surgical procedures, to guide future quality improvement initiatives. Methods: This retrospective cohort study analyzed 1322 patients who underwent select open pelvic and open abdominal surgeries between January 1 and December 31, 2017, in a tertiary health care centre in Montreal. Results: Patients who underwent open abdominal surgeries were prescribed significantly higher daily doses of morphine milligram equivalents (MME) (45 mg; interquartile range, 30-60), than patients who underwent either a caesarean delivery (20 mg, 20-20) or a hysterectomy (30 mg, 22-30). After adjustment for multiple potential confounders, abdominal surgery was associated with 4 times the odds of receiving more than 50 MME at hospital discharge compared with pelvic surgeries (odds ratio, 3.96; 95% confidence interval, 1.31-11.97). The availability of postoperative preprinted order sets with fixed high doses of opioids was also highly associated with the outcome. Conclusion: In our institution, some surgeries were more likely to receive high-risk opioid prescriptions at discharge. Efforts to optimize safer prescribing practices should address the creation and/or updating of preprinted order sets to reflect current best practice guidelines. This initiative could be overseen by hospital pharmacy and therapeutics committees.

OBJECTIVE: This study used a virtual patient simulation (VPS) to quantifiably and objectively assess undergraduate (UG) to postgraduate (PG) medical learners' acquisition of the entrustable professional activity (EPA) "handover," focusing particularly on the transition to residency. This EPA is critical because it is part of a core competency for UG and PG training in both the United States and Canada, and is essential for patient safety and comprehensive professional communication. DESIGN: Data were collected from 3 separate groups of participants: 2 UG cohorts from an earlier study, as well as a PG cohort at the beginning of residency. All participants completed the same trauma VPS, which required a free text summary statement that was used as a surrogate for an oral handover. These were collected and scored independently, using previously developed validated rubrics, one procedural and the second semantic. SETTING: All study participants were from one site. The VPS case was completed online. PARTICIPANTS: Two different UG groups, one designated junior (N = 52), was studied at the beginning of their clerkship year, a second group, designated senior (N = 30), was studied at the end of their clerkship year. These groups were compared to a third group of PG learners (N = 31) during the initial 2 weeks of their residency. Informed consent was obtained from all participants. RESULTS: A procedural rubric assessed learners' cognitive knowledge of trauma care-management. A semantic rubric assessed their use of the professional language necessary for a safe and succinct clinical handover communication. An Analysis of Variance comparing scores on the procedural rubric was highly significant with Tukey LSD tests indicating that all 3 groups were significantly different. Students increased their scores on the procedural rubric at each stage of their training. A parallel Analysis of Variance comparing students' scores on the semantic rubric revealed no significant increase in scores, indicating that students did not improve in their capacity to communicate professionally as they progressed through their training. CONCLUSIONS: Taken together, these results demonstrate that training was successful in teaching cognitive-based procedures, but not effective in teaching professional communication, which is critical to the EPA handover. Greater emphasis needs to be placed on ensuring the acquisition of professional communication skills throughout the continuum of UG and PG clinical activities. Faculty development should serve as a support to assist medical educators to address this requirement. These results also demonstrate that VPS with associated objective and validated rubrics can be used as an assessment methodology to quantifiably measure learner performance with respect to the EPA handover. A similar strategy should be considered across the UG and PG continuum for other EPAs and could form the nexus for further research. (C) 2019 Association of Program Directors in Surgery. Published by Elsevier Inc. All rights reserved.

The move of medical education in North America to a model based on teaching competencies rather than knowledge requires new methods of assessment.As part of a longer-term project to create objective and quantifiable measures for the learning of these competencies, we argue that the focus should be on the Entrustable Professional Activities (EPAs) that are associated with each competency.In this paper we present a theoretical constructivist analysis of one such EPA, that of patient handover, which is critical to patient safety.By applying an Activity Theory analysis to this EPA, we were able to identify, and thus quantify, key components of the EPA.Analysis of student performance on an online virtual patient simulation of a trauma case indicated that surgical clerkship students were able to correctly mention the majority of the procedures that expert surgeons indicated were most important, but only used less than a third of the professional terms that experts thought appropriate.This result points to a need to increase direct instruction in professional communication among medical learners, while demonstrating that Activity Theory provides an analysis that not only captures what is happening in the learning, but also can be used to develop objective and quantifiable assessment metrics.

CONTEXT: Undergraduate medical students at a large academic trauma center are required to manage a series of online virtual trauma patients as a mandatory exercise during their surgical rotation. PURPOSE: Clinical reasoning during undergraduate medical education can be difficult to assess. The purpose of the study was to determine whether we could use components of the students' virtual patient management to measure changes in their clinical reasoning over the course of the clerkship year. In order to accomplish this, we decided to determine if the use of scoring rubrics could change the traditional subjective assessment to a more objective evaluation. BASIC PROCEDURES: Two groups of students, one at the beginning of clerkship (Juniors) and one at the end of clerkship (Seniors), were chosen. Each group was given the same virtual patient case, a clinical scenario based on the Advanced Trauma Life Support (ATLS) Primary Trauma Survey, which had to be completed during their trauma rotation. The learner was required to make several key patient management choices based on their clinical reasoning, which would take them along different routes through the case. At the end of the case they had to create a summary report akin to sign-off. These summaries were graded independently by two domain "Experts" using a traditional subjective surgical approach to assessment and by two "Non-Experts" using two internally validated scoring rubrics. One rubric assessed procedural or domain knowledge (Procedural Rubric), while the other rubric highlighted semantic qualifiers (Semantic Rubric). Each of the rubrics was designed to reflect established components of clinical reasoning. Student's t-tests were used to compare the rubric scores for the two groups and Cohen's d was used to determine effect size. Kendall's was used to compare the difference between the two groups based on the "Expert's" subjective assessment. Inter-rater reliability (IRR) was determined using Cronbach's alpha. MAIN FINDINGS: The Seniors did better than the Juniors with respect to "Procedural" issues but not for "Semantic" issues using the rubrics as assessed by the "Non-Experts". The average Procedural rubric score for the Senior group was 59% 13% while for the junior group, it was 51% 12% ((t(80)) = 2.715; p = 0.008; Cohen's d = 1.53). The average Semantic rubric score for the Senior group was 31% +/- 15% while for the Junior group, it was 28% +/- 14% ((t(80)) = 1.010; p =.316, ns). There was no statistical difference in the marks given to the Senior versus Junior groups by the "Experts" (Kendall's = 0.182, p = 0.07). The IRR between the "Non-Experts" using the rubrics was higher than the IRR of the "Experts" using the traditional surgical approach to assessment. The Cronbach's alpha for the Procedural and Semantic rubrics was 0.94 and 0.97, respectively, indicating very high IRR. The correlation between the Procedural rubric scores and "Experts" assessment was approximately r = 0.78, and that between the Semantic rubric and the "Experts" assessment was roughly r = 0.66, indicating high concurrent validity for the Procedural rubric and moderately high validity for the Semantic rubric. PRINCIPLE CONCLUSION: Clinical reasoning, as measured by some of its "procedural" features, improves over the course of the clerkship year. Rubrics can be created to objectively assess the summary statement of an online interactive trauma VP for "procedural" issues but not for "semantic" issues. Using IRR as a measure, the quality of assessment is improved using the rubrics. The "Procedural" rubric appears to measure changes in clinical reasoning over the course of 3rd-year undergraduate clinical studies. (C) 2017 Association of Program Directors in Surgery. Published by Elsevier Inc. All rights reserved.)