Page 55 - AIH-1-2

Artificial Intelligence in Health AI in AD – Diagnosis and monitoring

driving, industrial automation, and the widespread use of to diagnose and monitor effectively. Prompt and accurate
cell phones. Recent developments have notably bolstered identification is indispensable for the effective management
the efficiency, accuracy, and productivity of AI-optimized of AD. However, conventional diagnostic techniques
workflows in the health-care industry. AI involves the tend to heighten variability in diagnosis and may cause
application of sophisticated computational algorithms delays in the initiation of therapy, relying heavily on
to simulate complex cognitive functions of humans, clinical judgment. A promising avenue toward achieving
accomplished through learning and adaptation to gathered objective, accurate, and rapid diagnostic procedures lies in
data. Over the past decades, there has been a marked increase the integration of AI-based technology, which holds the
in both research and application of AI in healthcare, with the potential to revolutionize AD management. This review
potential to completely transform the sector. 1-4 highlights the rationale behind the potential of AI in
Similar to AI, the field of dermatology has experienced completely transforming AD monitoring and diagnostic
rapid growth owing to advancements in technology and procedures. While acknowledging the efficiency of AI,
inventions. This evolution has brought dramatic changes it also emphasizes the importance of problem-solving
in the diagnosis and treatment of dermatological illnesses. and fostering teamwork. These initiatives are essential in
maximizing the benefit of AI in improving the precision
Computer algorithms have proven to be invaluable tools for and effectiveness of AD monitoring and diagnosis
dermatologists, particularly in diagnosing diseases such as procedures.
malignant melanoma. Dermatology boasts a vast archive
5
of clinical, dermatoscopic, and dermatopathological 2. Methodology
images, positioning it as a leader in the application of AI
in medicine. Therefore, having a basic understanding of In conducting a systematic literature review on the
AI becomes essential for designing and evaluating medical contemporary utilization of AI in diagnosing and
research in this area. Consequently, investigating the monitoring individuals with AD, we implemented a
potential uses of AI in dermatological practice becomes rigorous methodology for gathering pertinent articles
imperative. from prominent databases, including PubMed, Springer,
and Elsevier. The search utilized specific keywords, such
Atopic dermatitis (AD) is a common inflammatory as “Artificial intelligence,” “Machine learning,” “Deep
skin disease that affects a significant proportion of learning,” and “Atopic dermatitis.” Article selection
dermatology patients worldwide, estimated to affect 1 – prioritized peer-reviewed studies in dermatology that
6
2% of the global population. Notably, between the 1980s specifically examined the integration of AI in the diagnosis
and the early 2000s, there was a discernible global surge and monitoring of AD. Articles that failed to meet these
in the prevalence of AD, particularly pronounced among criteria or published in languages other than English
children under the age of five, with rates ranging from 10% were systematically excluded from the study. Following
to 16.5%. Individuals with AD exhibit a wide range of data extraction, we summarized the key findings, and
7,8
clinical symptoms, categorized into six distinct subtypes the synthesized information was then carefully compiled
based on their origin. Among these, the most prevalent into a comprehensive literature review, providing valuable
subtype is early-onset, early-resolving; nonetheless, insights into the current state of knowledge, addressing
recurrence is frequently observed, with less severe challenges, and advocating for collaboration across intra-,
symptoms than the original episode. The three main inter-, trans-, and multi-disciplinary domains to optimize
9,10
symptoms of AD – skin inflammation, compromised skin the benefits of AI in improving the accuracy of AD
barrier function, and persistent itching-exerta negative diagnosis.
influence on the lives of those afflicted, significantly
impacting their quality of life and level of satisfaction with 3. Principle of AI
therapy. 11-13 Moreover, patients’ adherence to treatment AI encompasses various computational subfields, including
procedures is severely hindered by these symptoms. 14-16 machine learning (ML) and natural language processing
Staphylococcus aureus colonization often arises as a result of (NLP), enabling computer systems to mimic human
skin barrier dysfunction, exacerbating the degradation of cognitive functions (Figure 1). At present, ML, where
the barrier function. 17,18 Notably, AD is linked to systemic computers anticipate data without explicit programming,
inflammatory conditions, such as metabolic syndrome stands as the frontier of AI advancement. Essentially,
and cardiovascular disease, despite not being as visually computers “learn” from data, offering analyses without
conspicuous as psoriasis. 19 explicit guidance on trait prioritization. Dermatologists
Given its broad spectrum of clinical symptoms, AD offer compelling examples, such as identifying melanomas
presents as an unexpected illness that can prove challenging from clinical images, predicting the effectiveness of
5

Volume 1 Issue 2 (2024) 49 doi: 10.36922/aih.2775