Artificial Intelligence in Hair and Nail Disorders

October 2022 | Volume 21 | Issue 10 | 1049 | Copyright © October 2022


Published online September 26, 2022

doi:10.36849/JDD.6519

Shishira R. Jartarkar MDa, Anant Patil MDb, Anna Waskiel-Burnat MDc, Lidia Rudnicka MDc, Michela Starace MDd, Stephan Grabbe MDe, Mohamad Goldust MDe

aDepartment of Dermatology, Vydehi Institute of Medical Sciences and Research Centre University - RGUHS, Bangalore, India
bDepartment of Pharmacology, DY Patil Medical College, Navi Mumbai, India
cDepartment of Dermatology, Medical University of Warsaw, Warsaw, Poland
dDivision of Dermatology, Department of Experimental, Diagnostic and Specialty Medicine, University of Bologna, Bologna, Italy
eDepartment of Dermatology, University Medical Center of the Johannes Gutenberg University, Mainz, Germany

Abstract
Artificial intelligence (AI), a field of computer science, aims at simulating human intelligence with computers. Though AI has surpassed dermatologists in skin cancer detection, it still lags behind various other specialties like radiologists in broader adoption. Newer AI applications are becoming increasingly accessible. AI plays a role in various areas, such as medical image recognition, auxiliary diagnosis, and drug research and development. Dermatology has a prime position in implementation of AI in medical research due to its larger clinical, dermoscopic, and histopathological image database. Hence, it is crucial to consider the potential and emerging role of AI in dermatology clinical practice. There are already studies focusing on various skin disorders like cancer, psoriasis, atopic dermatitis, etc. This article provides an overview of AI and its applications in hair and nail disorders at present and its future potential.

J Drugs Dermatol. 2022;21(10):1049-1052. doi:10.36849/JDD.6519

INTRODUCTION

Artificial intelligence (AI), simulated human intelligence possessed by machines, has emerged as a major frontier in computer science research. AI introduces a fundamental change in dermatological clinical practice, making it essential for every dermatologist to have a broad knowledge and understanding of AI.1 In clinical practice, the approach to dermatological disease has been evolving rapidly with the advent of newer technologies and interventions.2

Dermatology has taken a prime position in implementing AI in the medical field due to its large clinical, dermoscopical, and dermatopathological image database.2 As a visual field with a profound database, dermatology has seen some advanced progress in AI research, especially in design and the automated interpretation of medical images.3 This requires a basic understanding of AI.

It is crucial to consider the potential and emerging role of AI in dermatological clinical practice. There are already a number of studies on AI focusing on skin disorders – mainly skin cancers, psoriasis, and atopic dermatitis. Hence, in this review, we provide an overview of AI and its applications in hair and nail disorders. We also discuss the basic concepts, the emergence of AI, existing clinical applications, opportunities, and the challenges faced in the implementation of AI in routine dermatological clinical practice.

MATERIALS AND METHODS

Basic Concepts in Artificial Intelligence
AI is a computer science that is involved in creating programs that aim to reproduce human cognition and processes involved in data analysis.4 It is defined by the Association of Artificial Intelligence Advancement as: "The scientific understanding of mechanisms underlying thought and intelligent behavior and their embodiment in machines."5

Strong AI or artificial general intelligence confers a machine with a human level of intelligence that has the ability to learn by itself to carry out various tasks. However, we are far from having such complex machines in reality.6 Currently, we have access to only weak/narrow AI, which is designed to focus on a single goal.

Machine learning is a subset of AI that aims at teaching machines to learn tasks automatically by inferring patterns from the data.3 This learning method can be supervised, semi-supervised, or unsupervised. In supervised methods, the machine is fed with a database of problems along with the answers. Simply put, supervised methods learn from examples. This setup is limited by the accuracy of the training set. In unsupervised methods, the machine analyses the input data without a pre-defined answer,7 whereas the semi-supervised method is a hybrid approach that involves both labeled and unlabeled data.6