Perceptions of Artificial Intelligence Integration into Dermatology Clinical Practice: A Cross-Sectional Survey Study

February 2022 | Volume 21 | Issue 2 | Original Article | 135 | Copyright © February 2022


Published online January 31, 2022

doi:10.36849/JDD.6398

Chapman Wei MD,a Nagasai C. Adusumilli MBA,b Adam J. Friedman MD FAAD,b Vishal A. Patel MD FAAD FACMSb

aStaten Island University Hospital Northwell Health, Staten Island, NY
bThe George Washington School of Medicine and Health Sciences, Washington, DC

Abstract
Background: Artificial intelligence (AI) is a growing field in dermatology and has great potential for integration into clinical practice. Our objective was to assess the perceptions of artificial intelligence in dermatology practice.
Methods: An IRB-approved 18-question online survey was distributed by email. Patients were stratified by age to assess for statistical differences among perceptions.
Results: 90 respondents fully completed the survey. 54 (60.0%) respondents were slightly familiar with AI, and 73 (81.1%) respondents have not incorporated AI into their clinical practice. 27.8% of respondents perceived AI as superior to a human provider’s experience some of the time. 94.4% of respondents would at least use AI for certain scenarios. 65.6% of respondents believed that AI would help patients with analyzing and managing electronic health records. 38.9% respondents predict that AI will not decrease or increase the need for dermatologists. 51.6% of respondents felt that AI will at least somewhat enhance the dermatologists’ ability to screen skin lesions. The three dermatology areas that AI was perceived to most beneficial were malignant skin lesions, benign skin lesions, and pigmentation disorders. Age of respondents did not have a significant impact on the perceptions of AI.
Conclusion: Our results show that dermatologists surveyed were generally positive toward embracing AI integration into clinical practice. Further studies should be conducted to confirm these findings.

J Drugs Dermatol. 2022;21(2):135-140. doi:10.36849/JDD.6398

INTRODUCTION

Various types of artificial intelligence (AI), including machine learning and its subfield, deep neural network, have been successful and increasingly implemented in the healthcare field during the past decade due to the ease of access to large volumes of data to feed AI algorithms.1 While many tedious tasks can be automated, allowing the physician to focus on higher cognitive responsibilities, AI algorithms have improved substantially such that they can self-learn and be trained with high accuracy and speed using a set of training data. For example, AI algorithms using convolutional neural networks have been able to interpret cerebral vascular accidents from computed tomography images of real patient cases faster than radiologists.2 In dermatology, machine-based learning has been involved in imaging modalities, such as photography, dermoscopy, and confocal microscopy, with error rates less than 5%.3,4 Despite the volume of studies that highlight the hallmarks of AI, there are few peer-reviewed studies that assessed the perceptions of the disruptive technology among the dermatologic workforce.5-7 Thus, we distributed an IRB-approved non-incentivized survey among those subscribed to the Orlando Dermatology, Aesthetic and Surgical Conference (ODAC) listserv to assess how they perceive AI integration into clinical dermatology. Additionally, we sought to assess differences in perceptions of AI based on age given that most previous studies about physicians around technology adoption sampled people who learned to use digital technology during their adult life.8

MATERIALS AND METHODS

An 18-question anonymized, voluntary English survey was designed and distributed by email using SurveyMonkey® (www.surveymonkey.com, San Mateo, CA). This survey study was approved by the GWU Institutional Review Board (NCR191752). Additionally, this survey was externally validated by faculty members of the GW Medical Faculty Associates Department of Dermatology. The first three questions (Q1–Q3) assessed the responder’s demographics in their age, healthcare role, and practice setting. The remaining questions (Q4–Q18) focused on the perceptions of AI. Two questions (Q6, Q18) were open-ended and follow-up questions. The survey opened on April 13, 2021 and closed on May 14, 2021 (31 days). Based on respondents’ age, we stratified the survey respondents into two groups:
Close Menu