» Back to Table of Contents

Research Highlight

Telehealth and AI Guidance for Veterans with Skin Disease

The onset of the COVID-19 pandemic significantly affected VA healthcare, leading to cancelled or deferred in-person clinic appointments. While telehealth offered options for continuing care, teledermatology was surprisingly limited because VA’s most used modality – consultative asynchronous teledermatology – still required patients to travel to a VA care site for digital photography of their skin. Many patients and VA staff were unable or unwilling to continue this practice and asynchronous teledermatology encounters plummeted for several months in 2020, only recently recovering to pre-pandemic levels. To mitigate against such disruption and also possibly serve as preludes to or even substitutes for routine in-person dermatologic care, emerging home-based teledermatology and artificial intelligence(AI) technologies may offer patients and their primary care providers solutions for remotely accessing skin evaluations and care.

While VA has long championed home telehealth for remote disease monitoring, it has only recently developed and deployed formal platforms that allow patients to conveniently share visual information and to receive clinicians’ feedback, both of which are necessary for dermatologic care at home. VA Video Connect allows Veterans to use their mobile devices to have real-time video visits with clinicians, but its relatively high bandwidth requirements limit the resolution with which skin can be assessed and restrict its utility for dermatology. Emerging in 2019, My VA Images (MVAI) is a clinician-gated, patient-facing, asynchronous telehealth mobile web-based application or app that invites patients to share with VA clinicians high-resolution digital photos (and video) using their mobile devices at home. MVAI has a teledermatology-specific module that asks patients questions about their skin and overall health history and prompts them to list medications they use for their skin, before guiding them in taking photos of their skin problem. VA dermatologists or their designees use the Virtual Care Manager (VCM) app to invite patients to use MVAI, to review the patient-submitted history and images from MVAI, and finally to write both instructions to the patient and a clinical note for VA’s electronic health record. When it’s time to use the app or read the dermatologist’s instructions, patients receive email or secure text message notifications containing links to MVAI.

Veteran-facing apps such as MVAI may also be precursors to the future possibility that, before referral to dermatology, patients with skin problems (or their primary care providers) might share skin photos to be interpreted initially not by a remote dermatologist but instead by the computer residing on their own mobile devices.

Under controlled conditions, AI models have had some success in classifying images diagnostically following training on high- quality curated skin image databases. However, their effectiveness in classifying raw patient-submitted images, much less those from the Veteran population, is unclear. Our current research project, “Improving dermatology access by direct-to-patient teledermatology and computer-assisted diagnosis,” is attempting to address this gap.

In collaboration with Dr. Maria Wei at San Francisco VA Medical Center and University of California, San Francisco, our project, operating in the background and separate from clinical care, utilizes MVAI’s Veteran-submitted skin images to train and validate an AI-driven computer vision model to classify skin images by diagnoses. Among many considerations, training such a model requires that the images include the diversity of Veterans’ skin tones to increase the likelihood that the model will be relevant to as many VA patients as possible. Therefore, we selected VA facilities in Atlanta, Denver, and San Francisco, overseen by local site investigators Drs. Howa Yeung, Mayumi Fujita, and Amanda Twigg, respectively, as the focus of our research in part because these sites’ patient demographics reflect much of the diversity of Veterans’ skin types.

Training an AI model also requires harvesting a large number of MVAI images of various dermatologic diagnoses. To optimize the number of images as well as MVAI’s actual impact on access to care at our study sites, it is desirable for MVAI’s implementation to be as effective as possible, including successful adoption by clinicians and patients, and overcoming challenges that we previously documented during an earlier phase of MVAI’s deployment.^1,2 Therefore, in collaboration with VA researchers in Boston led by Dr. Allen Gifford and at Durham and Dallas led by Drs. Heather King and George Jackson, our project’s aims also include measuring VA’s and Veterans’ readiness for direct-to-patient care and users’ satisfaction with MVAI as well as measuring the impact that MVAI has on access to skin care at our three select VA facilities. Our operational partner, VA’s Office of Connected Care, has allowed our team to provide input into the app’s development and supported its use at research study sites; it has also created the analytic tools necessary to monitor MVAI’s usage. This research is still in progress, but some preliminary observations on patients’ readiness for direct- to-patient care can be made based on the project’s separate collaboration with VA’s Office of Strategic Planning and Analytics (OSPA).

First, the environment for digital health is dynamic and growing more favorable in VA. Between 2019 and 2023, OSPA’s annual Survey of Veteran Enrollees Use of Health Care indicated that the fraction of enrollees using the internet at least occasionally rose from 79 percent to 87 percent.3 From 2022 to 2023, the fraction of enrollees accessing the internet via a mobile device rose from 49 percent to 78 percent, suggesting that web-based apps designed for mobile device use, such as MVAI, are increasingly accessible to a large majority of Veterans. While younger Veterans are very well-positioned to take advantage of apps such as MVAI, even older Veterans moved in that direction, with 76 percent of enrollees 65 years of age and older accessing the internet in 2023, up from 71 percent in 2022.

Second, accompanying Veterans’ growing access to digital technologies and the internet, there is readiness to use telehealth from home to increase access to healthcare. As part of our collaboration, OSPA augmented its 2023 Survey of Veteran Enrollees to specifically query Veterans on their readiness to use their internet access and digital technologies to mediate their healthcare.³ Half of respondents reported that they check web pages at least 1 to 3 days a week to obtain health information, and 84 percent do so at some time, suggesting a familiarity with using the internet for health that may be an important precursor to using web-based apps such as MVAI. When asked for their willingness to do online telehealth tasks, 76 percent of enrollees reported they would be willing to fill out health forms on a computer or mobile device, and 71 percent were willing to use an app to track health measures. Age was a factor, with >87 percent of younger Veterans willing to fill out forms, but only 61 percent of those 65 or older willing to do so. Similarly, while >81 percent of younger Veterans were willing to use an app to track a health-related measure, only 57 percent of those 65 or older were willing to do so.

Third, some Veterans are wary of teleconsultation from home, and most are currently less willing to use AI for healthcare. While 66 percent of enrollees were willing to share their own health information as part of a virtual consultation, and 69 percent were willing to receive a medical opinion virtually, only 45 percent were willing to share health information with a computer that makes a decision without interacting with a provider, and only 43 percent would receive health information from such a computer.³ Only 30 percent or less of those aged 65 or older were willing to use a computer making a decision without provider interaction. These results suggest that VA still has work to do to improve Veterans’ comfort with telehealth from home and most Veterans today may not accept using AI alone for their healthcare. Our research project is further assessing Veterans’ specific willingness to use direct-to-patient teledermatology apps such as MVAI and to use AI models for their dermatology care.

Direct-to-patient care by telehealth apps such as MVAI, and by AI, may offer potential benefits to patients in improving both their temporal and geographic access to high quality skin care. Both these technologies are still evolving and will need to be rigorously evaluated to demonstrate their accuracy as well as efficacy in improving access and ultimately in improving clinical outcomes. However, an equally important component of these technologies’ effectiveness will depend on optimally implementing them in VA’s healthcare system, including acclimating VA staff and Veterans so that they will be willing and engaged participants in direct-to-patient care.

References

1. Peracca SB, et al., “Organizational Readiness for Patient-Facing Mobile Teledermatology to Care for Established Veteran Patients in the United States,” Telemed E Health 2022;29:72-80.
2. Peracca SB, et al., “Implementation of Direct-to-Patient Mobile Teledermatology in VA,” Journal of General Internal Medicine 2024;39:97-105.
3. Office of Strategic Planning and Analytics Department of Veterans Affairs, Survey of Veteran Enrollees' Health and Use of Health Care 2022 and 2023.