INTRODUCTION: Student-run free clinics (SRFCs) offer medical students a unique opportunity to develop their clinical, diagnostic, and social skills while providing care to medically underserved communities. This study aims to evaluate the value of SRFC involvement on students self-reported confidence in various clinical domains and satisfaction with their medical education. METHODS: We conducted a single-center retrospective pre-post assessment at an urban academic institution among second- to fourth-year medical students. We administered a 25-item questionnaire capturing the scope of clinic involvement and assessing self-reported confidence in multiple clinical domains following a one-year-long participation in student-run free clinics. RESULTS: Fifty-six students completed the survey. Participation in SRFCs significantly increased self-reported confidence in patient history-taking (p < 0.001), performing oral presentations (p < 0.001) and physical exams (p < 0.001). Students also reported significantly greater confidence in working with translators (p < 0.001) or as part of an interprofessional team (p < 0.001) and understanding the needs of the population served (p < 0.001). Students also found SRCs to significantly improve their confidence in preparedness for clerkships (p < 0.001). SRFC involvement can improve medical students confidence in their clinical and interpersonal skills and enhance preparedness for clerkships and working with diverse patient groups. CONCLUSION: SRFCs are a useful tool in the medical school curriculum that help bridge the gap between classroom learning and clinic and may encourage practice in medically underserved communities. SRFCs also integrate classroom material and clinical practice, although standardized evaluation metrics need to be developed. SRFCs should be incorporated as a learning experience by medical schools nationwide.

This prospective, single-centre cohort study aimed to evaluate the impact of a portable vision reading device, OrCam Read, on vision-related quality-of-life and independent functional status in patients with low vision due to stroke or brain tumours. Six patients with poor visual acuity or visual field defects due to a stroke or a brain tumour were enrolled at a U.S. Ophthalmology Department. Participants were trained to use OrCam Read and given a loaner device for the 1 month duration of the study. Various assessments, including daily function tests, the National Eye Institute Visual Function Questionnaire-25, and the 10-item neuro-ophthalmic supplement, were administered at the first and last visits. Patients experience with the device was evaluated with weekly telephone and end-of-study satisfaction surveys. The main outcome measures were the patient satisfaction with OrCam and the mean assessment scores between enrolment and final visits. The intervention with OrCam significantly improved patients ability to complete daily tasks and participants reported good satisfaction with the device. The results also show non-significant improvement with distant activities, dependency, and role difficulties. Our findings demonstrate the feasibility of studying vision-related quality-of-life using a portable vision device in this patient population and pave the way for a larger study to validate the results of this study.

PURPOSE: To determine whether microstructural retinal changes, tumor features, and apolipoprotein E (APOE) ε4 polymorphism are correlated with clinically detectable treatment-associated cognitive dysfunction (TACD) in patients with lower-grade gliomas. DESIGN: Cohort study. PARTICIPANTS AND CONTROLS: Sixteen patients with lower-grade glioma at a United States academic ophthalmology department between January 2021 and November 2023. Normal controls were recruited from convenient sampling. METHODS: Montreal Cognitive Assessment (MoCA) scores and retinal changes were assessed in 6-month intervals. Apolipoprotein E genotyping was performed, and tumor details were recorded. Partial least-squares discriminant (PLSD) model was established to evaluate the association between TACD with APOE genotype, ophthalmic, and tumor features. MAIN OUTCOME MEASURES: The main outcome measure was cognitive status as measured by the MoCA score and analyzed in relation to ophthalmic measurements, tumor features, and APOE genotype. RESULTS: Median time to first eye examination was 34 months (2-266) from tumor diagnosis and 23 months (0-246) from radiation. Nine patients (56%) had abnormal cognition (MoCA <26/30). Montreal Cognitive Assessment scores were significantly worse in patients with temporal (22 ± 7.2) than frontal lobe tumors (26 ± 3.1, P = 0.02) and those with oligodendrogliomas (22 ± 4.1) than astrocytomas (26 ± 3.6, = 0.02). Patients with TACD had significant radial peripapillary capillary density loss (45% ± 4.6) compared with those with normal cognition (49% ± 2.6, P = 0.02). A PLSD model correlated MoCA scores with retinal nerve fiber thickness, intraocular pressure, foveal avascular zone, best-corrected visual acuity, months since first diagnosis, and tumor pathology (oligodendroglioma or not). Using these features, the model identified patients with TACD with 77% accuracy. Apolipoprotein E genotyping showed: 2 ε2/ε3 (13%), 10 ε3/ε3 (63%), and 1 ε3/ε4 (6%). CONCLUSIONS: Retinal microstructural changes may serve as biomarkers for TACD in patients with lower-grade gliomas. Temporal lobe tumors and oligodendrogliomas may increase susceptibility to TACD. Utilization of retinal markers may enhance TACD diagnosis, progression monitoring, and inform management of lower-grade patients with glioma. A larger study with serial eye examinations is warranted to evaluate the role of APOE ε4 and develop a predictive model. FINANCIAL DISCLOSURES: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

PURPOSE: Automated machine learning (AutoML) has emerged as a novel tool for medical professionals lacking coding experience, enabling them to develop predictive models for treatment outcomes. This study evaluated the performance of AutoML tools in developing models predicting the success of pneumatic retinopexy (PR) in treatment of rhegmatogenous retinal detachment (RRD). These models were then compared with custom models created by machine learning (ML) experts. DESIGN: Retrospective multicenter study. PARTICIPANTS: Five hundred and thirty nine consecutive patients with primary RRD that underwent PR by a vitreoretinal fellow at 6 training hospitals between 2002 and 2022. METHODS: We used 2 AutoML platforms: MATLAB Classification Learner and Google Cloud AutoML. Additional models were developed by computer scientists. We included patient demographics and baseline characteristics, including lens and macula status, RRD size, number and location of breaks, presence of vitreous hemorrhage and lattice degeneration, and physicians experience. The dataset was split into a training (n = 483) and test set (n = 56). The training set, with a 2:1 success-to-failure ratio, was used to train the MATLAB models. Because Google Cloud AutoML requires a minimum of 1000 samples, the training set was tripled to create a new set with 1449 datapoints. Additionally, balanced datasets with a 1:1 success-to-failure ratio were created using Python. MAIN OUTCOME MEASURES: Single-procedure anatomic success rate, as predicted by the ML models. F2 scores and area under the receiver operating curve (AUROC) were used as primary metrics to compare models. RESULTS: The best performing AutoML model (F2 score: 0.85; AUROC: 0.90; MATLAB), showed comparable performance to the custom model (0.92, 0.86) when trained on the balanced datasets. However, training the AutoML model with imbalanced data yielded misleadingly high AUROC (0.81) despite low F2-score (0.2) and sensitivity (0.17). CONCLUSIONS: We demonstrated the feasibility of using AutoML as an accessible tool for medical professionals to develop models from clinical data. Such models can ultimately aid in the clinical decision-making, contributing to better patient outcomes. However, outcomes can be misleading or unreliable if used naively. Limitations exist, particularly if datasets contain missing variables or are highly imbalanced. Proper model selection and data preprocessing can improve the reliability of AutoML tools. FINANCIAL DISCLOSURES: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.