AI-Based Tool for Preliminary Diagnosis of Dermatological Manifestations

Abstract

Dermatological disorders affect millions of people worldwide, but effective and timely diagnosis continues to be a challenge, particularly in underserved or rural areas with limited access to a dermatologist. Using deep learning techniques this study demonstrates an AI-enabled tool that will assist effective and timely diagnosis of skin conditions and will assist health care providers in making initial assessments. The tool analyzes images of skin ailments and provides the user a preliminary assessment. This skin assessment system will assist health professionals to make quicker and better decisions regarding diagnosing skin disorders and treating dermatological diseases. The purpose of the AI tool for dermatology is to increase efficiency of care, decrease time to treatment, and ultimately improve care for dermatopathy patients. The system will be improved by developing semi-automated location-based services to assist users in finding nearby hospitals or clinics. It will also make it more relevant and easier to use in the real-world contexts.

Introduction

Skin disorders are highly prevalent worldwide, and early diagnosis is essential to prevent disease progression and reduce healthcare burden. However, limited access to dermatologists—particularly in rural and remote areas—often delays diagnosis and treatment. Artificial Intelligence (AI), especially through Convolutional Neural Networks (CNNs), offers a promising solution by enabling rapid and accurate analysis of skin images to support preliminary diagnosis and improve access to dermatological care.

The text reviews recent advancements in AI-driven dermatology, highlighting how deep learning models outperform traditional machine learning and, in some cases, even match or exceed dermatologist-level diagnostic accuracy. CNN-based architectures such as VGG, ResNet, DenseNet, and hybrid models, along with techniques like transfer learning, GAN-based data augmentation, and multi-model fusion, have shown strong performance on public datasets such as ISIC, HAM10000, and DermNet. Despite these advances, challenges remain, including lack of explainability (black-box behavior), data imbalance across skin types, scarcity of high-quality labeled images, sensitivity to image quality, and concerns about bias and trust in clinical settings. The literature strongly emphasizes the need for Explainable AI (XAI), diverse datasets, lightweight models for edge deployment, and integration with clinical workflows.

The proposed system is an AI-enabled dermatological tool designed to provide fast, preliminary diagnosis using skin images. It employs a hybrid CNN–BiLSTM architecture optimized using the Hunger-Based Fruit Bat Optimization (HFBO) algorithm. The system processes images through data collection, preprocessing, feature extraction, classification, and optimization stages. It classifies skin lesions into four categories: Normal, Melanocytic Nevus, Melanoma, and Benign Keratosis. The tool includes a user-friendly interface that allows image upload or capture, provides diagnostic results with causes, precautions, and medication suggestions, and offers nearby clinic recommendations.

Experimental results demonstrate strong performance, with overall accuracy ranging from 94–96%, precision around 93%, recall 92%, and F1-score between 92–94%. The system delivers results within seconds, making it suitable for real-time, point-of-care screening. While limitations such as data quality and generalizability persist, the study concludes that AI-based dermatological tools can effectively support early screening, reduce diagnostic delays, and improve access to dermatological care, particularly in underserved regions.

Conclusion

This study offers a new AI tool that detects skin diseases, based on skin images provided by patients. The hybrid CNN-BiLSTM model was developed to steadily, and accurately classify skin conditions including melanoma, nevus, acne, and benign keratosis. Additionally, the AI tool is expected to enhance patient engagement and accurately detect skin conditions, while advising on disease (safety and treatment approaches). The user engagement will be further enhanced by the user-friendly activity and the use of \"locate\" for locating nearby hospitals- it makes the user feel like they are being guided to a facility. Overall, the research offers a new tool for increasing screening skin cancer rates, helping patients make decisions about their treatment options and therefore screening patients more efficiently to reduce the burden on the dermatologist.

References

[1] Vamshi Yadav, Shashank Moudgalya C P, Srikanth Gowda N, Rishi N Lekkala, Dr. Kuppala Saritha, “AI BASED TOOL FOR PRELIMINARY DIAGNOSIS OF DERMATOLOGICAL MANIFESTATIONS”, 2025. [2] SANDHIYA DEVI M, SHANJANASREE P, SONA S, MS. SWETHA, “AI BASED TOOL FOR PRELIMINARY DIAGNOSIS OF DERMATOLOGICAL MANIFESTATIONS,” May 2025. [3] Ling-Fang Li Xu Wang Wei-Jian Hu Neal N. Xiong Yong-Xing Du Bao-Shan Li, “Deep Learning in Skin Disease Image Recognition”, 2020-11-01. [4] Nourah Janbi Rashid Mehmood Iyad Katib Aiiad Albeshri Juan M. Corchado Tan Yigitcanlar, “Imtidad: A Reference Architecture and a Case Study on Developing Distributed AI Services for Skin Disease Diagnosis over Cloud, Fog and Edge”, 2022. [5] Zhouxiao Li, Konstantin Christoph Koban, Thilo Ludwig Schenck, Riccardo Enzo Giunta, Qingfeng Li, and Yangbai Sun, “Artificial Intelligence in Dermatology Image Analysis: Current Developments and Future Trends”, 2022-11-18. [6] Prof. SR Kale, Mr. Mahesh Dandekar, Mr. Mahesh Bawane, Mr. Parth Dangre, Mr. Parth Sabale, “AI BASED TOOL FOR PRELIMINARY DIAGNOSIS OF DERMATOLOGICAL MANIFESTATIONS”, 2025-04-01 [7] Yali Nie, Paolo Sommella, Marco Carratù, Matteo Ferro, Mattias O’Nils, and Jan Lundgren, “Recent Advances in Diagnosis of Skin Lesions Using Dermoscopic Images Based on Deep Learning”, 2022-08-17. [8] Prof. S. R. Kale, Parth V. Sable, Parth S. Dangre, Mahesh R. Dandeker, Mahesh Bawane, “A Review on AI-Based Tool for Preliminary Diagnosis of Dermatological Manifestations”, January–February 2025. [9] Rashmi N. Kuppast, Preetham R, K. Chandan, Chethan K, “AI TOOL FOR PRELIMINARY DIAGNOSIS OF DERMATOLOGY MANIFESTATION”, 2024 (Vol-10, Issue-3, IJARIIE Journal). [10] Dr. Chandrasekar Vadivelraju, Keren Elisheba S, Vikhyath Moodbidri Bharath Kumar, Pavaman Suraj, Merugu Harish Reddy, “Dermascan AI: Deep Learning System for Preliminary Diagnosis of Dermatological Manifestations”, 2025 (Volume 13, Issue 5, May 2025). [11] Prof. S. R. Kale, Parth V. Sable, Parth S. Dangre, Mahesh R. Dandeker, and Mahesh Bawane., “A Review on AI-Based Tool for Preliminary Diagnosis of Dermatological Manifestations”, 2025. [12] Zhouxiao Li, Konstantin Christoph Koban, Thilo Ludwig Schenck, Riccardo Enzo Giunta, Qingfeng Li, and Yangbai Sun, “Artificial Intelligence in Dermatology Image Analysis: Current Developments and Future Trends\"”. [13] Yali Nie, Paolo Sommella, Marco Carratù, Matteo Ferro, Mattias O’Nils, and Jan Lundgren, “Recent Advances in Diagnosis of Skin Lesions Using Dermoscopic Images Based on Deep Learning”, 2022-08-17.

Copyright

Copyright © 2026 Junaid Ahmed N, Ms. Kalpana K Harish, Mohammad Sadiq Badashasaheb Banakar, Indrajeet Madiwal. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.