IMPORTANCEThere was a shift in patient volume from in-person to video telemedicine visits during the COVID-19 pandemic. OBJECTIVE To determine the concordance of provisional diagnoses established at a video telemedicine visit with diagnoses established at an in-person visit for patients presenting with a new clinical problem. DESIGN, SETTING, AND PARTICIPANTS This is a diagnostic study of patients who underwent a video telemedicine consultation followed by an in-person outpatient visit for the same clinical problem in the same specialty within a 90-day window. The provisional diagnosis made during the video telemedicine visit was compared with the reference standard diagnosis by 2 blinded, independent medical reviewers. A multivariate logistic regression model was used to determine factors significantly related to diagnostic concordance. The study was conducted at a large academic integrated multispecialty health care institution (Mayo Clinic locations in Rochester, Minnesota; Scottsdale and Phoenix, Arizona; and Jacksonville, Florida; and Mayo Clinic Health System locations in Iowa, Wisconsin, and Minnesota) between March 24 and June 24, 2020. Participants included Mayo Clinic patients residing in the US without age restriction. Data analysis was performed from December 2020 to June 2021. EXPOSURES New clinical problem assessed via video telemedicine visit to home using Zoom Care Anyplace integrated into Epic. MAIN OUTCOMES AND MEASURES Concordance of provisional diagnoses established over video telemedicine visits compared against a reference standard diagnosis. RESULTS There were 2393 participants in the analysis. The median (IQR) age of patients was 53 (37-64) years; 1381 (57.7%) identified as female, and 1012 (42.3%) identified as male. Overall, the provisional diagnosis established over video telemedicine visit was concordant with the in-person reference standard diagnosis in 2080 of 2393 cases (86.9%; 95% CI, 85.6%-88.3%). Diagnostic concordance by International Statistical Classification of Diseases and Related Health Problems, TenthRevision chapter ranged from 64.7% (95% CI, 42.0%-87.4%) for diseases of the ear and mastoid process to 96.8% (95% CI, 94.7%-98.8%) for neoplasms. Diagnostic concordance by medical specialty ranged from 77.3% (95% CI, 64.9%-89.7%) for otorhinolaryngology to 96.0% (92.1%-99.8%) for psychiatry. Specialty care was found to be significantly more likely than primary (continued) Key Points Question How concordant to an in-person diagnosis are provisional diagnoses established at a video telemedicine visit for patients presenting with a new clinical problem? Findings In this diagnostic study of 2393 patients who underwent a video telemedicine consultation followed by an in-person outpatient visit for the same clinical problem in the same specialty within a 90-day window, the provisional diagnosis established over video telemedicine visit matched the in-person reference standard diagnosis in 86.9% of cases. Meaning These findings suggest that video telemedicine visits yield a high deg...

Frozen sections are a useful pathologic tool, but variable image quality may impede the use of artificial intelligence and machine learning in their interpretation. We aimed to identify the current research on machine learning models trained or tested on frozen section images. We searched PubMed and Web of Science for articles presenting new machine learning models published in any year. Eighteen papers met all inclusion criteria. All papers presented at least one novel model trained or tested on frozen section images. Overall, convolutional neural networks tended to have the best performance. When physicians were able to view the output of the model, they tended to perform better than either the model or physicians alone at the tested task. Models trained on frozen sections performed well when tested on other slide preparations, but models trained on only formalin‐fixed tissue performed significantly worse across other modalities. This suggests not only that machine learning can be applied to frozen section image processing, but also use of frozen section images may increase model generalizability. Additionally, expert physicians working in concert with artificial intelligence may be the future of frozen section histopathology.

No abstract

No abstract