An Automated Approach for Diagnosing Allergic Contact Dermatitis Using Deep Learning to Support Democratization of Patch Testing

Mayo Clinic Proceedings. Digital health Pub Date : 2024-02-23 DOI:10.1016/j.mcpdig.2024.01.006

Matthew R. Hall MD , Alexander D. Weston PhD , Mikolaj A. Wieczorek BA , Misty M. Hobbs MD , Maria A. Caruso BA , Habeeba Siddiqui BA , Laura M. Pacheco-Spann MS , Johanny L. Lopez-Dominguez MD , Coralle Escoda-Diaz BA , Rickey E. Carter PhD , Charles J. Bruce MB, ChB

{"title":"An Automated Approach for Diagnosing Allergic Contact Dermatitis Using Deep Learning to Support Democratization of Patch Testing","authors":"Matthew R. Hall MD , Alexander D. Weston PhD , Mikolaj A. Wieczorek BA , Misty M. Hobbs MD , Maria A. Caruso BA , Habeeba Siddiqui BA , Laura M. Pacheco-Spann MS , Johanny L. Lopez-Dominguez MD , Coralle Escoda-Diaz BA , Rickey E. Carter PhD , Charles J. Bruce MB, ChB","doi":"10.1016/j.mcpdig.2024.01.006","DOIUrl":null,"url":null,"abstract":"<div><h3>Objective</h3><p>To develop a deep learning algorithm for the analysis of patch testing.</p></div><div><h3>Patients and Methods</h3><p>A retrospective case series between January 1, 2010, and December 31, 2020, was constructed to develop a deep learning model for the classification of patch test results from photographs. The performance of human expert readers reviewing the same photographs blinded to the original clinical physical examination findings was measured to benchmark model performance.</p></div><div><h3>Results</h3><p>Model performance on the independent test set (n=5070 test site locations from 37 patients) achieved an area under the receiver operating characteristic curve of 0.89 (95% CI, 0.86-0.91) and an F1 score of 37.1. The optimal cutoff had a sensitivity of 70.1% (136/194; 95% CI, 63.1%-76.5%) and a specificity of 91.7% (4472/4876; 95% CI, 90.9%-92.5%).</p></div><div><h3>Conclusion</h3><p>We demonstrated proof-of-concept utility for detecting allergic contact dermatitis using an automated deep learning approach.</p></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"2 1","pages":"Pages 131-138"},"PeriodicalIF":0.0000,"publicationDate":"2024-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2949761224000075/pdfft?md5=ec00a6cd5a159970ea4a21e054923ea6&pid=1-s2.0-S2949761224000075-main.pdf","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Mayo Clinic Proceedings. Digital health","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2949761224000075","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}

引用次数: 0

Abstract

Objective

To develop a deep learning algorithm for the analysis of patch testing.

Patients and Methods

A retrospective case series between January 1, 2010, and December 31, 2020, was constructed to develop a deep learning model for the classification of patch test results from photographs. The performance of human expert readers reviewing the same photographs blinded to the original clinical physical examination findings was measured to benchmark model performance.

Results

Model performance on the independent test set (n=5070 test site locations from 37 patients) achieved an area under the receiver operating characteristic curve of 0.89 (95% CI, 0.86-0.91) and an F1 score of 37.1. The optimal cutoff had a sensitivity of 70.1% (136/194; 95% CI, 63.1%-76.5%) and a specificity of 91.7% (4472/4876; 95% CI, 90.9%-92.5%).

Conclusion

We demonstrated proof-of-concept utility for detecting allergic contact dermatitis using an automated deep learning approach.

查看原文本刊更多论文

利用深度学习支持斑贴测试民主化的过敏性接触性皮炎自动诊断方法

目标开发一种用于分析斑贴测试的深度学习算法。患者与方法构建了2010年1月1日至2020年12月31日期间的回顾性病例系列，以开发一种深度学习模型，用于对照片中的斑贴测试结果进行分类。结果模型在独立测试集（来自 37 名患者的 5070 个测试部位）上的表现达到了接收器操作特征曲线下面积 0.89（95% CI，0.86-0.91）和 F1 分数 37.1。最佳临界值的灵敏度为 70.1%（136/194；95% CI，63.1%-76.5%），特异性为 91.7%（4472/4876；95% CI，90.9%-92.5%）。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Mayo Clinic Proceedings. Digital health Medicine and Dentistry (General), Health Informatics, Public Health and Health Policy

自引率

0.00%

发文量

审稿时长

47 days