Commercialization of Deep Learning Skin Diagnosis Model… Expanding Application in Wearable and Medical Fields

Deep Learning-Based AI for Skin Irritation Diagnosis... Innovative Technology with 98.3% Accuracy Emerges

New artificial intelligence diagnosis technology that automatically detects skin reactions has emerged, marking an important turning point in the cosmetics research field. This technology transitions from traditional methods of manual observation to AI that proves its potential by diagnosing skin irritation reactions quickly and accurately.

Data-Driven Transformation in Skin Diagnosis Environment

The most challenging aspect of evaluating skin irritation is the varied standards of irritation perceived by individuals. Even the same irritation can be assessed slightly differently by experts, leading to discrepancies based on experience. Considering these factors, the emergence of AI technology inevitably becomes a solution to ensure accuracy and objectivity in skin diagnosis.

Article Source

A recently developed AI-based automatic skin irritation diagnostic technology by a research team offers a fundamental solution to the limitations of existing diagnostic systems. This model, which has learned from a total of 83,629 actual skin irritation data instances, faithfully reflects the structure of the patch test performed by experts. The trained AI autonomously assigns irritation scores from 0 to 4 and categorizes reactions with accuracy similar to human judgment.

"The strength of a data-driven approach lies in the fact that AI trained on expert standards can assess skin conditions consistently."

The fact that skin irritation diagnosis is being transformed from a mere sensory issue to a scientific and systematic analysis is an encouraging sign.

Object Detection Analysis Model Implemented with YOLOv5x

The core algorithm of this AI diagnostic technology is YOLOv5x. Recognized as one of the latest object detection models, this technology is optimized for quickly and accurately detecting specific objects or phenomena in images. The research team customized and applied this algorithm to suit skin irritation diagnosis.

The subjects are images taken immediately after the patch test. This structure involves analyzing the skin reaction area 24 and 48 hours post-test to evaluate the degree of irritation. Consequently, this method proved effective by achieving a high accuracy of 98.3% at both time points.

Additionally, the fact that AI correctly identified an 'irritation 0' state, or non-irritation state, with a sensitivity of 99.7% is considered a critical reliability indicator.

"The detection algorithm of YOLOv5x can capture even very minor changes in the image, allowing for high precision separation of skin irritation reactions."

This technology goes beyond the precision of image-based diagnosis by also securing efficiency over time.

Reliability of the Technology Based on Experimental Verification

The research team utilized extensive evaluation data to demonstrate the performance of the AI model. By separating and repeatedly testing a total of 1,312 evaluation data and 1,536 verification data, the systematic verification of the model's generalization performance was achieved.

Most importantly, this technology is not merely a prototype. The official academic journal of the European Society of Contact Dermatitis (ESCD), ‘Contact Dermatitis’, published the research results online, granting both academic and industrial validation of the technology.

"Publishing the AI-based diagnostic technology in an international academic journal signifies securing the scientific foundation of the technology while opening its global applicability."

In summary, this not only produces accurate results but provides a reliable benchmark that could transform the paradigm of skin diagnosis.

Changes in the Skin Care Paradigm Brought About by Technology Expansion

The most significant advantages of AI irritation diagnostic technology are ‘consistency’, ‘speed’, and ‘data-driven decision-making’. For the first time, a pathway has opened for quantifying the diagnosis of skin conditions, which can appear differently to each individual.

In particular, compared to traditional methods that require considerable testing personnel and time, automated AI technology can process large amounts of data quickly while maintaining the same quality. This structure offers flexibility, making it applicable in the cosmetics industry as well as in medical, research, and clinical trial fields.

"The irritation reaction determined by AI can be utilized in various ways such as constant monitoring or trend analysis, transforming the customer experience itself."

Ultimately, the more accurate the diagnosis, the more precise the actual skin-tailored prescriptions become. Product development based on individual skin reaction data is no longer a dream.

Possibility of Follow-up Research and Industrial Application

This technology continues to evolve. One direction is to develop more advanced diagnosis models by designing algorithms differently according to maximum skin reaction time or various test areas. Additionally, training on the unique reactions of elderly or children's skin, or patients with specific skin diseases, can provide specialized diagnostics, which is a feasible scenario.

---

YURICO5 provides a more efficient and accurate research environment as a trusted cosmetics research partner through AI-based skin irritation diagnostic technology. Based on this, with YURICO5, new possibilities in skin science can be continuously explored.