Comprehensive Summary
In response to the need for telemedicine appointments in the wake of the COVID-19 pandemic, Vodrahalli et al. examined the ability of AI to improve telemedicine photo quality. In this study, adults aged 18 and older who presented to clinic with a skin condition were given a smartphone with an AI algorithm installed, providing them with the ability to take a photo of their skin. After taking a photo, the AI algorithm provided each patient with feedback on the photo quality; if insufficient, AI would instruct the patient to re-take the photo until the image was sufficient (maximum of 4 photo attempts). Results of this study showed that the AI algorithm reduced the number of patients with poor-quality images by 68%. The most common reasons for poor photo quality included blurs and poor lighting. Additionally, across all patients, the algorithm was able to significantly improve the photo quality, resulting in an easier identification of skin lesion images and clinical diagnosis. These results suggest that AI algorithms can be used to identify weaknesses in patient-provided images, improve their quality for later clinical use and, importantly, diagnostic purposes.
Outcomes and Implications
The use of AI algorithms to improve patient-provided images shows a new step for telemedicine, signaling a future avenue for diagnostic medicine in the field of dermatology. If we can improve the quality of patient-provided images, more personalized care is possible from the comfort of the patient’s home, which is highly important in the case of people located in remote areas who are still in need of dermatologic care but may not be able to access it. Additionally, with improved image quality, dermatologists will have an easier time diagnosing and providing subsequent care. With this advancement, apps on mobile devices can begin to be used in a widespread manner, increasing access for groups who need it most.