ObjectivesTo evaluate the uptake of a platform for virtual visits in primary care, examine patient and physician preferences for virtual communication methods and report on characteristics of visits and patients experience of care.DesignA retrospective cohort study.SettingPrimary care practices within five regions in Ontario, Canada after 18 months of access to virtual care services.Participants326 primary care providers and 14 291 registered patients.InterventionsProviders used a platform that allowed them to connect with their patients through synchronous (audio/video) and/or asynchronous (secure messaging) communication.Main outcome measuresUser-level data from the platforms including patient demographics, practice characteristics, communication modality used, visit characteristics and patients’ satisfaction.ResultsAmong the participants, 44% of registered patients and 60% of registered providers used the platform at least once. Among patient users, 51% completed at least one virtual visit. The majority of virtual visits (94%) involved secure messaging. The most common patient requests were for medication prescriptions (24%) and follow-up from previous appointment (22%). The most common provider request was to follow-up on test results (59%). Providers indicated that 81% of virtual visits required no follow-up for that issue and 99% of patients reported that they would use virtual care services again.ConclusionsWhile there are a growing number of primary care video visit services, our study found that both patients and providers in rostered practices prefer secure messaging over video. Despite fears that virtual visits would be overused by patients, when patients connected with their own primary care provider, many virtual visits appeared to replace in-person visits, and patients did not overwhelm physicians with requests. This approach may improve access and continuity in primary care.

An estimated 3 billion people lack access to dermatological care globally. Artificial intelligence (AI) may aid in triaging skin diseases and identifying malignancies. However, most AI models have not been assessed on images of diverse skin tones or uncommon diseases. Thus, we created the Diverse Dermatology Images (DDI) dataset—the first publicly available, expertly curated, and pathologically confirmed image dataset with diverse skin tones. We show that state-of-the-art dermatology AI models exhibit substantial limitations on the DDI dataset, particularly on dark skin tones and uncommon diseases. We find that dermatologists, who often label AI datasets, also perform worse on images of dark skin tones and uncommon diseases. Fine-tuning AI models on the DDI images closes the performance gap between light and dark skin tones. These findings identify important weaknesses and biases in dermatology AI that should be addressed for reliable application to diverse patients and diseases.

No abstract

No abstract

No abstract