Potential for the development of customized color cosmetics using artificial intelligence digital slide scanner exclusively for dermatology: a scoping review

Introduction

As the 4th industrial revolution is taking place worldwide, artificial intelligence (AI) and machine learning (ML) have been playing a major role in the medical field for the past few years. This has also made great progress in breast pathology (1). The development of AI is leading to various changes in the medical field. The use of AI in the medical field is very diverse. It is possible to confirm biomarker quantification, derive Ki67 and cell division count, and evaluate lymph node metastasis in normal and frozen sections (2). It also performs grade and surrounding stroma evaluation (3). It is possible to check general matters such as treatment response and prognosis prediction (4). In depth, it is known that it is possible to even detailed applications such as quantification of tumor-infiltrating lymphocytes and diagnosis of breast lesions in frozen sections (5). As digital health becomes more widespread; it has become essential to modern practice. Virtual slides are becoming an important clinical, educational, and research tool in pathology. These virtual microscopes digitize glass slides by collecting hundreds of tiles of areas of interest at various magnification levels. These gigapixel images are the same as how a pathologist uses a microscope (6). Digital pathology (DP) is increasingly being used in cancer diagnosis. It provides a new tool for faster, higher quality and more accurate diagnosis. The practice of this diagnostic pathology supports computational histopathology and AI-based diagnosis. DP can provide digital images to enhance and enhance and can use advanced AI algorithms and computer-aided diagnosis techniques. This use of AI has brought about tremendous changes in the way these new tools are used (7). Digital slide scanners have a variety of applications, especially in dermatology (8). They are of great help in the diagnosis of skin cancer. Whole body scanners allow dermatologists to take high-resolution images of the entire patient’s body to monitor skin lesions. They can also be used to track changes in skin cancer over time. Advances in slide scanner technology have allowed large numbers of tissue slides to be scanned. This vast amount of data is now being represented and stored digitally. In DP, this has many uses. It could have a significant impact on remote pathology consultation and education (8,9). This new source of “big data” also presents tremendous research opportunities in the field of image computing. It can leverage data that contains basic prognostic data. This can extract “sub-visual” image features that pathologists cannot visually identify in DP slide images. The ability to mine these features can be used to better quantitatively model disease emergence and comparative analysis (8). This provides an opportunity to improve predictions of disease aggressiveness and patient outcomes (10).

Recently, in Korea, due to the rapid transition to a non-face-to-face society after coronavirus disease-19 (COVID-19), consumers are becoming more aware of non-face-to-face treatment without visiting medical institutions. This has led to the birth of the Direct-To-Consumer concept, where consumers can receive genetic testing directly. In addition, based on the 4th industrial revolution, consumers have increased their needs for customized inner beauty products and customized cosmetics (11). Beauty has a great impact on an individual’s self-esteem and social interactions, and healthy skin is a key element of beauty (12,13). Various skin diseases have a high prevalence worldwide and have a great impact on the quality of life. Early diagnosis and management are important for serious diseases such as skin cancer (14,15). In addition, makeup not only improves appearance, but also plays a role in protecting and managing the skin as part of skin health management. Appropriate skin care and cosmetic use are essential for maintaining skin health. Therefore, customized cosmetics tailored to an individual’s skin condition and preferences should optimize skin health and provide effective management methods that meet individual needs. Advances in AI and digital technology enable the development of such customized products (16,17). Accordingly, the research results confirm that a convergence medical device is needed to secure the function of antioxidants as NRF2 regulators (11). In this way, we can confirm the emergence of convergence medical devices based on the 4th industrial revolution that reflected consumer needs and the need for customized dermatological cosmetics (18).

The advancement of AI is also bringing about revolutionary changes in aesthetic dermatology. Previous studies have mainly emphasized the high performance of AI-based diagnostic support, especially in skin cancer diagnosis. However, the problems of subjective evaluation and non-standardized methods have been pointed out in the application of aesthetic dermatology. Studies by Koski et al. (19) and Mudgal et al. (20) emphasize the need for standardized evaluation methods and diverse dataset collection to solve these problems (21). However, studies on the medical potential of AI for personalized cosmetics and the issues of fairness and bias are lacking. This study aims to bridge this knowledge gap and suggest new ethical considerations in the field of dermatology by exploring the medical potential of personalized cosmetics using AI and a digital slide scanner.

Therefore, this study aims to investigate the medical potential of personalized cosmetics that maintain healthy beauty by utilizing AI and digital slide scanners. In line with the era of AI healthcare, we aim to develop personalized cosmetics in the field of dermatology by providing data to improve personal skin health and beauty. We present this article in accordance with the PRISMA-ScR reporting checklist (available at https://jmai.amegroups.com/article/view/10.21037/jmai-24-372/rc).

Methods

Accordingly, relevant keywords were individually set and performed. The search method is as follows: we searched PubMed, Medline, Scopus, ResearchGate, and Google Scholar using ‘healthcare’, ‘artificial intelligence’, ‘AI health’, ‘AI’, ‘personalized cosmetics’, ‘color-customized cosmetics’, ‘digital slide scanner’, ‘AI digital slide scanner’, and ‘dermatology digital slide scanner’. This review was conducted as a technical review according to the PRISMA-ScR guidelines. A total of 46 articles were finally included. The study model and selection process are presented in Figures 1,2. Accordingly, a total of 2,721 documents were searched. In the final stage, 46 documents were selected and included in the paper. This is indicated as PRISMA-ScR in Figure 1. Figure 1 systematically shows the flowchart of the process of finding and selecting studies in this scoping review. In addition, a model diagram of the studies is shown in Figure 2. Figure 3 shows the country distribution of the included studies.

Figure 1 PRISMA flow diagram.

Figure 2 Research model diagram. AI, artificial intelligence.

Figure 3 Country distribution of included studies (N=46).

Search strategy

This study used PubMed, Medline, Scopus, ResearchGate, and Google Scholar databases. The search period was from July 1 to 14, 2024. The language was restricted to English, and the study type was restricted to review, research, and experimental studies. The data included in this study were limited to a 5-year period from 2019 to 2024 to reflect recent research. A total of 46 articles were included in the reviewed databases (n=2,043) and included registers (n=678). Records removed prior to screening included content not in English (n=1,123), records deemed ineligible by the automated tool (n=450), records removed for other reasons (n=438), screened records (n=710), and excluded records (n=640). We deleted 640 articles that had the same keywords but differed from the content of this study. Therefore, a total of 46 articles were included, including studies in the review (n=33) and reviews of studies (n=13). This is represented as a PRISMA flow diagram in Figure 1. This study covers the latest advances in the application of AI technology in the medical field, with a particular emphasis on image quality improvement models. It also examines the potential of AI-based cosmetic development and the use of AI pigments for dermatologists. This is to explore the relationship between dermatologists and AI, and to evaluate the potential of developing dyeing cosmetics. The research model is included in Figure 2.

Study selection

This study was conducted according to the following process. Two researchers established a clear process. They systematically reviewed the studies. First, they checked whether the studies included quantitative data. Next, they checked the title first. Then, they reviewed whether the content was appropriate. In addition, they excluded studies that only had an abstract or that had no content. Finally, they reviewed the abstracts of the articles and assessed all studies separately according to the inclusion criteria. The core criteria included only studies related to the content of this study, deleting studies that differed from the content of the affiliated study. This iteration was repeated by two researchers to ensure that the assessment was clear. Therefore, 46 publications were included in this study. Studies that were irrelevant to this study, inapplicable, or had no abstract or text were excluded from the final items.

Data extraction and management

This study extracted data using a standardized data extraction form. It also included outcome measures. The search algorithm is as follows: [(‘healthcare’ OR ‘artificial intelligence’ OR ‘personalized inner beauty’ OR ‘AI health’ OR ‘personalized cosmetics’ OR ‘color-customized cosmetics’ OR ‘digital slide scanner’ OR ‘AI digital slide scanner’ OR ‘dermatology digital slide scanner’ OR ‘personalized health’) AND (dermatology with artificial intelligence)]. To clarify the extraction form, PRISMA-ScR was performed on this study to standardize data collection. The data were added to this study. To address this, the research search strategy is specified in Tables 1,2. Additionally, two researchers separated the extracted data into locations. A multiple review method was used. The main references of the 46 studies included in this study were summarized, and the characteristics of the selected studies were presented in Table 3. The final number of selected studies was 46. They were published in 15 countries, most of which were published in the Republic of Korea (N=16), followed by the United States (N=12). This is shown in Table 4 as the percentage of countries in this paper’s references (N=46). Figure 3 shows the country’s distribution of the included studies (N=46).

Results

AI technology being used in the medical field is summarized in Result 1. Applications of AI in Medical Applications are summarized in Result 2. The latest technology in image quality improvement models using AI is summarized in Result 3. Cosmetics development through AI’s visual understanding is summarized in Result 4. The research model diagram according to this is shown in Figure 2. Therefore, the potential for the development of customized color cosmetics using AI digital slide scanner exclusively for dermatology suggests infinite development potential.

AI technology being used in the medical field

AI technologies are being used in a variety of ways in healthcare. The potential value of AI in nursing, in healthcare, ranges from improving the quality and efficiency of care to delivering on the promise of personalized and precise medicine. As more and more data accumulate about all aspects of health, AI systems may become virtually indispensable. AI can help reduce variability in care while increasing accuracy, accelerating discovery, and bridging gaps in healthcare (19). AI should not be viewed as a potential source of future 22nd-century “digital tyranny”. To provide a balanced view of the potential and challenges of AI in healthcare, several specifics are discussed and utilized. AI and related technologies are advancing rapidly. These advances will enable healthcare providers to create new value for patients and improve operational efficiency (20). Furthermore, AI may ultimately prove beneficial to healthcare. However, careful governance like physician behavior governance is needed to ensure sustainable safety. Regulatory guidance on how to safely implement and evaluate AI technologies is needed. Further research is needed on the specific capabilities and limitations of medical use of AI technology (21).

Applications for AI in medical applications

AI can also analyze medical images. It analyzes medical images such as X-rays, MRIs, and CT scans. This image analysis is used to detect lesions in patients and help diagnose them. This allows doctors to make more accurate judgments. Recent studies have shown that as radiation therapy technology advances, it is utilized for safe high-dose delivery. Better imaging technology is essential in the future. The development of AI for image-guided radiation therapy technologies ranging from kilovoltage and megavoltage methods to two- and three-dimensional technologies has been discussed in depth (22,23). In addition, the use of medical question-answering AI models can be mentioned. Medical question-answering AI models help answer medical questions. For example, Google Health’s Med-PaLM 2 provides up-to-date information in the medical field. It can also be confirmed that it provides accurate answers to medical questions. As such, AI technology has been attracting continuous attention in various fields including health recently. This is an innovation in the medical field. It started with the increase in computer hardware and software applications and the digitization of health-related data. The development and use of AI in the medical field is accelerating. These advances bring new opportunities and challenges (24). It suggests the future direction of AI in the health field. In addition, the major data types identified in the new research areas were multi-omics, clinical, behavioral, environmental, and pharmaceutical research and development (R&D) data. The status of AI related to each data type is described. These studies describe the related challenges and practical implications that have emerged in the past few years (24,25).

The latest technology in image quality improvement models using AI

The latest technology of image quality improvement models using AI, and the research results that AI can contribute to image quality improvement and the ability to interpret it are being confirmed. Recently, it has been confirmed that deep neural network-based research is actively being conducted in the field of image quality improvement. This study utilizes various optimization techniques and convolutional neural networks (CNNs). Through this, the lighting components in the image are estimated. This explores methods for improving image quality (26). To solve problems that exist in scene recognition research, we are conducting multi-faceted research. We study a new CNN target detection model to achieve a better balance between the accuracy and speed of high-speed scene image recognition. In this research problem, the following problems are identified. First, to solve the problem that the image is easily disturbed, and the quality is poor due to impurities in detailed image recognition, a preprocessing method based on a Canny edge detector is designed. Through this, the Canny operator is introduced to process gray images. Second, the L2 regularization algorithm optimizes the basic network framework of the CNN. This improves the stability of the model in complex environments. It also improves the generalization ability of the model. It is also used to improve the recognition accuracy of the algorithm to a certain level. Compared to the basic CNN algorithm, the algorithm is confirmed to have better recognition performance. In addition, it has been shown to have excellent generalization ability (26). This can also be confirmed in the illumination component estimation and improvement method and additional research. The pixel value in the image is expressed as the product of the illumination component and the reflection component. In this study, the illumination component is estimated to be based on Retinex theory1. This can be improved through gamma correction, etc. An optimization method for simultaneous estimation of the illumination component and the reflection component can be confirmed. Through this study, it is confirmed that the image quality improvement performance has been improved. Recently, in imaging medical technology, the computer’s information processing ability and resource storage ability have greatly improved, and neural network technology is also supported. CNNs have excellent characterization functions in computer vision tasks such as image recognition technology. It can confirm the image recognition and classification problem with high similarity in a specific field. This proposes a high similarity image recognition and classification algorithm that combines CNNs (27). In addition, changes can be confirmed in CNN-based image quality improvement research. CNNs can recognize images. They have shown excellent performance in object detection, etc. By utilizing this, the potential features of the image can be effectively detected. Methods for generating such improved images are being studied. In the medical AI field, a hybrid CNN-long short-term memory (LSTM) review for diabetes prediction was conducted. This is a CNN-LSTM method that combines CNN and LSTM. This study showed that it has better diabetes prediction performance than other deep learning methods. The CNN-LSTM model can identify hidden features between physiological variables. It also showed good performance in extracting correlations. Therefore, it is confirmed that it can be used to predict diabetes. Like other deep neural network architectures, the CNNLSTM model is said to have better detection accuracy when using a large data set (28).

Cosmetics development through AI’s visual understanding

3D Scene Understanding Product development is being done as an example of AI’s visual understanding and application through object recognition and understanding in 3D space. LG Household & Health Care in Korea has developed and is selling the Mini Tattoo Printer ‘IMPRINT’ on the market. It can be confirmed that this is an innovative product that reflects the needs of consumers rather than simply bringing in images. This mini tattoo printer is a portable printer that allows customers to select a tattoo design on a mobile app and implement it on their body. It is small enough to hold in one hand, light in weight, and uses HP’s cartridge technology to enable clear printing. This printer uses ‘Vegan Ink for Skin Makeup’ developed by LG Household & Health Care’s Color Research Institute, allowing you to enjoy tattoos harmless to your skin. The tattoo lasts for about 24 hours. It can also be easily erased by washing with a body cleanser, allowing you to express your individuality by improvising a tattoo you like anytime, anywhere (29). Geometric semantic scene understanding is an essential spatial intelligence function for robots to perceive and explore the world. However, it is still difficult for robots to understand natural scenes due to limited sensors and situations that change over time (30,31). We propose a novel scene understanding approach that seamlessly captures geometric and semantic aspects in the environment to bridge the gap between robots and animal perception. In addition, to verify the practicality of the proposed scene understanding method, studies on real geometric semantic scene reconstruction in indoor environments using a self-developed drone are also being conducted (31,32).

Discussion

Main findings

This scoping review is the first study to highlight the potential use of customized color cosmetics using dermatological AI and digital slide scanner as a countermeasure to maintain healthy beauty in the AI healthcare era to improve skin. The AI healthcare era has confirmed the medical possibility and necessity of customized color cosmetics using AI digital slide scanner for dermatology by utilizing big data and holographic images using massive data. However, there has been no research based on human security that can promote sustainability in the AI healthcare era. In addition, there has been no prior research on the development of AI and digital slide scanner and hologram devices utilizing the 4th industrial revolution for the image of healthy skin. Therefore, it is thought that it will be of great help in the development of products that consider the image of healthy human skin and healthy beauty in the AI healthcare era. Therefore, further research on this is necessary.

What does customized cosmetics exclusively for dermatologists have to do with AI?

A lot of research is being conducted on personalized healthcare using big data, and it is being practically applied and utilized in the medical industry (26,27). It is also a reality that it is being applied to the cosmetics industry. South Korea’s LG Household & Health Care and Amore Pacific cosmetics companies are also utilizing it in the field of customized cosmetics (33,34). They are developing customized cosmetics that consider the skin condition and preferences of individuals, and for this purpose, they are presenting personalized skin solutions by utilizing genetic analysis and precise skin measurement (35,36). It is believed that these efforts are intended to reflect the diverse needs of consumers and provide more customized products in the beauty market (36). It is expected that in the future, as AI develops, it will be utilized in the development of various products that reflect consumer needs (37,38).

Is it possible to use a digital slide scanner for AI skin health?

A study was conducted to evaluate two whole slide imaging applications in dermatopathology. This used a digital slide approach. It was compared to the gold standard traditional microscope using glass slides. The main diagnosis was inflammatory, no melanotic and melanotic lesions. Digital slides were captured using a commercially available scanner. The time taken to examine each slide was summarized as follows. The average time per slide was 23 seconds for the microscope, 34 seconds for the in-house application and 38 seconds for the vendor application. The results confirm that the digital slide viewing application is not yet efficient for evaluating dermatopathology cases. It also emphasizes the need for future improvements in the capture process and presentation of digital slides (39). DP image analysis using digital slide scanners is being widely conducted. There is also a high demand for decision support systems based on this. In the paper on staining normalization of histopathology images, a study was conducted based on nonlinear mapping of the source image to the target image using a representation derived from color deconvolution. This explains how color deconvolution affects color. It is a method to obtain staining intensity values when a staining matrix is given. It also relies on a standard staining matrix that may be inappropriate for the given image. We newly propose a color-based classifier that incorporates a novel staining color descriptor to compute such image-specific staining matrices. Therefore, the color normalization paradigm as a preprocessing step is confirmed to be generally related to imaging conditions in histological image analysis algorithms. It suggests that it can be of great help in demonstrating stable performance that is not sensitive to scanner changes. However, based on these research results, computerized decision support systems require changes in tissue preparation. In addition, since many problems occur due to color changes in tissue shape caused by using scanners from different manufacturers, additional research is needed (40). In the metaverse world that has changed to a non-face-to-face era after COVID-19, there is a need to technically verify the customer experience of consumers in the cosmetics market (13,14). In the dermatology cosmetics market, we must continue to change the problems of users who test and purchase cosmetics in person so that we can use the metaverse for new marketing and customer experiences in a non-face-to-face society (17,41).

What does this study add to existing knowledge, and how does this review differ from previous reviews?

Unlike previous reviews that primarily discuss current trends and applications of AI in aesthetic dermatology (42), such as Thunga et al. (43) and Kania et al. (44), this study provides a comprehensive scoping review using the PRISMA-ScR guidelines. It integrates a wide range of keywords and databases to provide a more detailed and systematic analysis. This study highlights the medical potential and necessity of personalized cosmetics utilizing AI and digital slide scanners for skin health, which has not been extensively addressed in previous reviews. It also highlights the importance of fairness and bias in AI systems, providing a nuanced perspective on the ethical considerations needed for future development. By addressing the need for further research on fairness and bias in AI systems, this study lays the foundation for more inclusive and equitable advancements in AI-based dermatology.

Limitations of this study

This study systematically emphasizes the potential of customized color cosmetics devices for dermatology using AI and digital slide scanners as a response to sustainable beauty in the AI healthcare era. However, there are several limitations that this study could not resolve. First, there is a limitation that the number of papers related to AI and dermatology is small. In addition, there is a lack of evidence to support customized cosmetics. However, this result is reflected in the fact that no research has been published yet on color cosmetics using AI and digital slide scanners. Second, it has not been confirmed as a research topic related to the AI healthcare era and color cosmetics, and further research is needed. In addition, there is a lack of research on the applicability and necessity of AI and customized color devices. Third, there is a lack of research on bias and fairness related to AI ethics. This will require deep reinforcement learning and 3D scene understanding, which are the foundations of AI that thinks and judges on its own, by conducting research on the fairness and bias of AI systems. In addition, research on the latest AI pre-technology, such as inference using ultra-large AI language models and bias and fairness related to AI ethics, is expected to be important. This is expected to be a major study that can solve the homework of human security. Therefore, the researcher plans to conduct follow-up research to reflect the fairness and bias of AI systems to consumers’ needs (45,46).

Conclusions

This scoping review investigated the medical potential of customized color cosmetics using AI for dermatology and digital slide scanner as a measure to maintain healthy beauty in the AI healthcare era. The results of this study suggest that the world is moving toward the AI healthcare era due to the development of medicine through AI, and that healthy beauty is needed in response. In addition, the development and use of AI and digital slide scanner and holograms are continuing in the field of dermatology medicine. This suggests the possibility of utilizing not only customized management using massive information but also personal color using digital slide scanner. This suggests a new area of AI healthcare from the perspective of sustainable human security. This suggests that there is infinite potential in the development of customized color cosmetics and skin health management using AI for dermatology and digital slide scanner in the future.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the PRISMA-ScR reporting checklist. Available at https://jmai.amegroups.com/article/view/10.21037/jmai-24-372/rc

Peer Review File: Available at https://jmai.amegroups.com/article/view/10.21037/jmai-24-372/prf

Funding: None.

Conflicts of Interest: Both authors have completed the ICMJE uniform disclosure form (available at https://jmai.amegroups.com/article/view/10.21037/jmai-24-372/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Yousif M, van Diest PJ, Laurinavicius A, et al. Artificial intelligence applied to breast pathology. Virchows Arch 2022;480:191-209. [Crossref] [PubMed]
2. Badve SS. Artificial intelligence in breast pathology - dawn of a new era. NPJ Breast Cancer 2023;9:5. [Crossref] [PubMed]
3. Jose L, Liu S, Russo C, et al. Generative Adversarial Networks in Digital Pathology and Histopathological Image Processing: A Review. J Pathol Inform 2021;12:43. [Crossref] [PubMed]
4. Campanella G, Hanna MG, Geneslaw L, et al. Clinical-grade computational pathology using weakly supervised deep learning on whole slide images. Nat Med 2019;25:1301-9. [Crossref] [PubMed]
5. Al Nemer AM. Application of artificial intelligence in the field of breast pathology diagnosis: narrative review. J Med Artif Intell 2024;7:37. [Crossref]
6. Cooper M, Ji Z, Krishnan RG. Machine learning in computational histopathology: Challenges and opportunities. Genes Chromosomes Cancer 2023;62:540-56. [Crossref] [PubMed]
7. Shafi S, Parwani AV. Artificial intelligence in diagnostic pathology. Diagn Pathol 2023;18:109. [Crossref] [PubMed]
8. Burns BL, Rhoads DD, Misra A. The Use of Machine Learning for Image Analysis Artificial Intelligence in Clinical Microbiology. J Clin Microbiol 2023;61:e0233621. [Crossref] [PubMed]
9. Jiang S, Guo C, Song P, et al. High-throughput digital pathology via a handheld, multiplexed, and AI-powered ptychographic whole slide scanner. Lab Chip 2022;22:2657-70. [Crossref] [PubMed]
10. Jaunay EL, Dhillon VS, Semple SJ, et al. Can a digital slide scanner and viewing technique assist the visual scoring for the cytokinesis-block micronucleus cytome assay? Mutagenesis 2020;35:311-8. [Crossref] [PubMed]
11. Lee J, Kwon KH. Development of customized inner beauty products and customized cosmetics apps according to the use of NRF2 through DTC genetic testing after the COVID-19 pandemic. J Cosmet Dermatol 2022;21:2288-97. [Crossref] [PubMed]
12. Lee J, Kwon KH. DTC genetic test for customized cosmetics in COVID-19 pandemic: Focused on women in their 40s and 60s in Seoul, Republic of Korea. J Cosmet Dermatol 2021;20:3085-92. [Crossref] [PubMed]
13. Lee J, Kwon KH. Recognition and development of customized cosmetics for military trainees in 20s and 30s in Republic of Korea. Health Sci Rep 2021;4:e334. [Crossref] [PubMed]
14. Lee J, Kwon KH. Recognition and the development potential of mobile shopping of customized cosmetic on untact coronavirus disease 2019 period: Focused on 40’s to 60’s women in Seoul, Republic of Korea. J Cosmet Dermatol 2021;20:1975-91. [Crossref] [PubMed]
15. Lee J, Kwon KH. Skin problems of Korean military personnel changes in the use of cosmetics and differences in preference according to different characteristics: Focused on comparison pre- and post-enlistment. Health Sci Rep 2021;4:e368. [Crossref] [PubMed]
16. Lee J, Kwon KH. Changes in the use of cosmetics worldwide due to increased use of masks in the coronavirus disease-19 pandemic. J Cosmet Dermatol 2022;21:2708-12. [Crossref] [PubMed]
17. Lee J, Kwon KH. Artificial intelligence technologies for color matching cosmetics in dermatology: a systematic review focusing on OCA2 gene direct-to-consumer genetic testing. J Med Artif Intell 2025;8:8. [Crossref]
18. Lee J, Kwon KH. Sustainable changes in beauty market trends focused on the perspective of safety in the post-coronavirus disease-19 period. J Cosmet Dermatol 2022;21:2700-7. [Crossref] [PubMed]
19. Koski E, Murphy J. AI in Healthcare. Stud Health Technol Inform 2021;284:295-9. [PubMed]
20. Mudgal SK, Agarwal R, Chaturvedi J, et al. Real-world application, challenges and implication of artificial intelligence in healthcare: an essay. Pan Afr Med J 2022;43:3. [PubMed]
21. Aung YYM, Wong DCS, Ting DSW. The promise of artificial intelligence: a review of the opportunities and challenges of artificial intelligence in healthcare. Br Med Bull 2021;139:4-15. [Crossref] [PubMed]
22. Niu T, Tsui T, Zhao W. AI-Augmented Images for X-Ray Guiding Radiation Therapy Delivery. Semin Radiat Oncol 2022;32:365-76. [Crossref] [PubMed]
23. Vliegenthart R, Fouras A, Jacobs C, et al. Innovations in thoracic imaging: CT, radiomics, AI and x-ray velocimetry. Respirology 2022;27:818-33. [Crossref] [PubMed]
24. Wang F, Preininger A. AI in Health: State of the Art, Challenges, and Future Directions. Yearb Med Inform 2019;28:16-26. [Crossref] [PubMed]
25. Sallam M. ChatGPT Utility in Healthcare Education, Research, and Practice: Systematic Review on the Promising Perspectives and Valid Concerns. Healthcare (Basel) 2023;11:887. [Crossref] [PubMed]
26. Wang P, Qiao J, Liu N. An Improved Convolutional Neural Network-Based Scene Image Recognition Method. Comput Intell Neurosci 2022;2022:3464984. [Crossref] [PubMed]
27. Liu Z, Sun L, Zhang Q. High Similarity Image Recognition and Classification Algorithm Based on Convolutional Neural Network. Comput Intell Neurosci 2022;2022:2836486. [Crossref] [PubMed]
28. Soltanizadeh S, Naghibi SS. Hybrid CNN-LSTM for Predicting Diabetes: A Review. Curr Diabetes Rev 2024;20:e201023222410. [Crossref] [PubMed]
29. Jang I. LG Household & Health Care to Unveil Mini Tattoo Printer ‘Imprintu’ at CES 2024. Health Trends 2024. Available online: https://www.k-health.com/news/articleView.html?idxno=69539
30. An Y. LG Household & Health Care Mini Tattoo Printer ‘Imprintu’ at ‘Urban Break 2023’. THE K BEAUTY SCIENCE 2023. Available online: https://www.thekbs.co.kr/news/articleView.html?idxno=10784
31. Zhang C, Yang Z, Xue B, et al. Perceiving like a Bat: Hierarchical 3D Geometric-Semantic Scene Understanding Inspired by a Biomimetic Mechanism. Biomimetics (Basel) 2023;8:436. [Crossref] [PubMed]
32. Nakaura T, Naganawa S. Writing medical papers using large-scale language models: a perspective from the Japanese Journal of Radiology. Jpn J Radiol 2023;41:457-8. [Crossref] [PubMed]
33. Lee TL, Ding J, Trivedi HM, et al. Understanding Radiological Journal Views and Policies on Large Language Models in Academic Writing. J Am Coll Radiol 2024;21:678-82. [Crossref] [PubMed]
34. Lee J, Kwon KH. A cross-sectional study on the use of big data for the past H1N1 influenza epidemic in obesity after COVID-19: Focused on the body slimming cream and leptin via DTC gene test. J Cosmet Dermatol 2022;21:5321-35. [Crossref] [PubMed]
35. Lee J, Kwon KH. The significant transformation of life into health and beauty in metaverse era. J Cosmet Dermatol 2022;21:6575-83. [Crossref] [PubMed]
36. Lee J, Kwon KH. Future perspective safe cosmetics: Focused on associated with ISO natural organic index. J Cosmet Dermatol 2022;21:6619-27. [Crossref] [PubMed]
37. Lee J, Kwon KH. Changes and expectations of the digitally underprivileged in artificial intelligence: a systematic review focusing on skin health for the welfare of the elderly. J Med Artif Intell 2024;7:30. [Crossref]
38. Lee J, Kwon KH. Skin health response to climate change weather tailored cosmetics using artificial intelligence. J Med Artif Intell 2024;7:20. [Crossref]
39. Cazzato G, Rongioletti F. Artificial intelligence in dermatopathology: Updates, strengths, and challenges. Clin Dermatol 2024;42:437-42. [Crossref] [PubMed]
40. Sauter D, Lodde G, Nensa F, et al. Deep learning in computational dermatopathology of melanoma: A technical systematic literature review. Comput Biol Med 2023;163:107083. [Crossref] [PubMed]
41. Lee J, Kwon KH. Future value and direction of cosmetics in the era of metaverse. J Cosmet Dermatol 2022;21:4176-83. [Crossref] [PubMed]
42. Elder A, Cappelli MO, Ring C, et al. Artificial intelligence in cosmetic dermatology: An update on current trends. Clin Dermatol 2024;42:216-20. [Crossref] [PubMed]
43. Thunga S, Khan M, Cho SI, et al. AI in Aesthetic/Cosmetic Dermatology: Current and Future. J Cosmet Dermatol 2025;24:e16640. [Crossref] [PubMed]
44. Kania B, Montecinos K, Goldberg DJ. Artificial intelligence in cosmetic dermatology. J Cosmet Dermatol 2024;23:3305-11. [Crossref] [PubMed]
45. Giovanola B, Tiribelli S. Beyond bias and discrimination: redefining the AI ethics principle of fairness in healthcare machine-learning algorithms. AI Soc 2023;38:549-63. [Crossref] [PubMed]
46. Yang Y, Lin M, Zhao H, et al. A survey of recent methods for addressing AI fairness and bias in biomedicine. J Biomed Inform 2024;154:104646. [Crossref] [PubMed]