BT18: Acceptability of artificial intelligence among dermatologists

Rashi Pangti,¹ Sanjeev Gupta,² Praanjal Gupta,³ Ambika Dixit,⁴ Hem Sati Chandra⁵ and Somesh Gupta⁵

¹Lady Hardinge Medical College and Associated Hospitals, New Delhi, India; ²Maharishi Markandeshwar Institute of Medical Sciences and Research, Ambala, India; ³JIPMER, Pondicherry, India; ⁴Deen Dayal Upadhaya College, New Delhi, India; and ⁵All India Institute of Medical Sciences, New Delhi, India

The field of medicine is generally slow to adopt newer technology. There is often apprehension among clinicians as to how this will affect their practice. Artificial intelligence (AI) is coming up in medicine, more so in specialties that depend on an image-based diagnosis like radiology, histopathology and dermatology. In dermatology, several machine learning algorithms for diagnosis have been developed and are freely available for use. We undertook this study to establish the acceptability of AI among dermatologists, their attitude towards it and specific apprehensions associated with it. We prepared a questionnaire using Google Forms and circulated it among qualified dermatologists and dermatology trainees. The questionnaire comprised of 12 statements, the reply to which was one of the following: strongly agree; agree; neither agree nor disagree; disagree; strongly disagree. There were also four questions, the reply to which was ‘yes’, ‘no’ or ‘can’t say’. There were 166 respondents (99 males, 67 females). The mean (SD) age of respondents was 36·45 (13) years. The mean (SD) duration of experience was 7·80 (10·92) years. There was no difference in the perception of AI-based on public or private sector employment or between the two sexes. A greater percentage of older (> 35 years) than younger (≤ 35 years) dermatologists perceived that AI will benefit dermatology more than other medical specialties (P = 0·03) and that any new development in AI in dermatology is welcome (P = 0·012). Younger dermatologists perceived more often that AI may replace dermatologists in the future (P = 0·007). Dermatologists with more than 5 years’ experience agreed that AI should be made part of training during dermatology residency (P = 0·001) and were more interested in AI (P = 0·004) than the trainees and those with less than 5 years’ experience. Most (62·05%) dermatologists responded that they were interested in AI. The majority also perceived that AI will be used more by general practitioners (75·3%), alternative medicine practitioners, traditional healers (79·5%) and by patients for self-diagnosis and self-treatment (84·3%). A greater percentage of older than younger dermatologists were open to using AI. This may reflect greater apprehension among younger dermatologists who are yet to settle in their career after completing their degree. Overall, there is a positive attitude towards AI among most dermatologists seen in our study. Thus, our study brings a fresh perspective to these specific apprehensions among dermatologists. This can then be taken care of by the regulatory bodies involved in AI in healthcare.