{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,5]],"date-time":"2026-06-05T14:16:15Z","timestamp":1780668975573,"version":"3.54.1"},"reference-count":20,"publisher":"BMJ","issue":"1","license":[{"start":{"date-parts":[[2026,3,27]],"date-time":"2026-03-27T00:00:00Z","timestamp":1774569600000},"content-version":"unspecified","delay-in-days":26,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc\/4.0\/"}],"content-domain":{"domain":["bmj.com"],"crossmark-restriction":true},"short-container-title":["BMJ Health Care Inform"],"accepted":{"date-parts":[[2026,3,3]]},"published-print":{"date-parts":[[2026,3]]},"abstract":"<jats:sec>\n                    <jats:title>Objective<\/jats:title>\n                    <jats:p>The study aims to develop an artificial intelligence (AI) framework for automatic pressure injury (PI) staging directly from raw clinical images, without requiring manual lesion localisation. By integrating a two-stage deep learning approach\u2014combining object detection and image classification\u2014the proposed system seeks to improve diagnostic accuracy, reduce subjectivity and support consistent staging in real-world clinical workflows.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>We conducted a retrospective study using 1807 PI images from China Medical University Hospital collected between 2020 and 2024. Lesions were annotated by five senior nurses with consensus. YOLOv9 was used for lesion detection, and DenseNet161 for staging. Performance metrics included accuracy, sensitivity, specificity, F1 score, area under the curve (AUC) and mean average precision (mAP)@0.5.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>The object detection model (YOLOv9) achieved an mAP@0.5 of 0.796. On the independent test set (n=365), the staging model demonstrated an overall accuracy of 0.775, sensitivity of 0.775, specificity of 0.955, precision of 0.779 and F1 score of 0.775.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion<\/jats:title>\n                    <jats:p>The proposed two-stage AI framework\u2014using YOLOv9 for lesion localisation and DenseNet161 for staging\u2014demonstrated performance comparable to or exceeding previously reported approaches, while offering improved clinical interpretability when applied to real-world clinical images. By separating detection and classification, the framework reduced background noise and improved the ability to distinguish visually similar PI stages. However, challenges remain in handling intra-wound heterogeneity, subtle early-stage boundaries and variability in image quality.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>This two-stage AI system may assist clinical staff by providing standardised, reproducible PI staging from clinical images, with potential for integration into nursing workflows.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1136\/bmjhci-2025-101636","type":"journal-article","created":{"date-parts":[[2026,3,27]],"date-time":"2026-03-27T15:40:17Z","timestamp":1774626017000},"page":"e101636","update-policy":"https:\/\/doi.org\/10.1136\/crossmarkpolicy","source":"Crossref","is-referenced-by-count":1,"title":["Novel two-stage deep learning framework for automated pressure injury classification"],"prefix":"10.1136","volume":"33","author":[{"ORCID":"https:\/\/orcid.org\/0009-0007-3561-0211","authenticated-orcid":false,"given":"Ting-Yu","family":"Lai","sequence":"first","affiliation":[{"name":"Artificial Intelligence and Robotic Innovation Center, China Medical University Hospital, Taichung, Taiwan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7271-2566","authenticated-orcid":false,"given":"Yi-Jiun","family":"Chou","sequence":"additional","affiliation":[{"name":"Department of Nursing, China Medical University Hospital, Taichung, Taiwan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0006-0611-9143","authenticated-orcid":false,"given":"Chun-You","family":"Liu","sequence":"additional","affiliation":[{"name":"Artificial Intelligence and Robotic Innovation Center, China Medical University Hospital, Taichung, Taiwan"},{"name":"Department of Biomedical Informatics, China Medical University, Taichung, Taiwan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0009-3429-0062","authenticated-orcid":false,"given":"Chien-Wei","family":"Chen","sequence":"additional","affiliation":[{"name":"Artificial Intelligence and Robotic Innovation Center, China Medical University Hospital, Taichung, Taiwan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6925-9999","authenticated-orcid":false,"given":"Ching-Ting","family":"Lin","sequence":"additional","affiliation":[{"name":"Artificial Intelligence and Robotic Innovation Center, China Medical University Hospital, Taichung, Taiwan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2239-8174","authenticated-orcid":false,"given":"Wei-Chun","family":"Wang","sequence":"additional","affiliation":[{"name":"Artificial Intelligence and Robotic Innovation Center, China Medical University Hospital, Taichung, Taiwan"},{"name":"Department of Neurology, China Medical University Hospital, Taichung, Taiwan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9821-3746","authenticated-orcid":false,"given":"Yimin","family":"Hsu","sequence":"additional","affiliation":[{"name":"Department of Nursing, China Medical University Hospital, Taichung, Taiwan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0968-0551","authenticated-orcid":false,"given":"Ming-Li","family":"Hsieh","sequence":"additional","affiliation":[{"name":"Department of Nursing, China Medical University Hospital, Taichung, Taiwan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1664-4818","authenticated-orcid":false,"given":"Shih-Sheng","family":"Chang","sequence":"additional","affiliation":[{"name":"Artificial Intelligence and Robotic Innovation Center, China Medical University Hospital, Taichung, Taiwan"},{"name":"Division of Cardiovascular Medicine, China Medical University Hospital, Taichung, Taiwan"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"239","published-online":{"date-parts":[[2026,3,27]]},"reference":[{"key":"2026032708401080000_33.1.e101636.1","doi-asserted-by":"crossref","first-page":"157","DOI":"10.1016\/j.jtv.2017.04.003","article-title":"The effect of pressure and shear on tissue viability of human skin in relation to the development of pressure ulcers: a systematic review","volume":"26","author":"Hoogendoorn","year":"2017","journal-title":"J Tissue Viability"},{"key":"2026032708401080000_33.1.e101636.2","doi-asserted-by":"publisher","DOI":"10.1097\/WON.0000000000000281"},{"key":"2026032708401080000_33.1.e101636.3","doi-asserted-by":"crossref","first-page":"773","DOI":"10.1016\/j.pmr.2022.06.002","article-title":"Comprehensive Management of Pressure Injury: A Review","volume":"33","author":"Kandi","year":"2022","journal-title":"Phys Med Rehabil Clin N Am"},{"key":"2026032708401080000_33.1.e101636.4","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijnurstu.2020.103546"},{"key":"2026032708401080000_33.1.e101636.5","doi-asserted-by":"crossref","first-page":"105266","DOI":"10.1016\/j.ijmedinf.2023.105266","article-title":"Augmented Decision-Making in wound Care: Evaluating the clinical utility of a Deep-Learning model for pressure injury staging","volume":"180","author":"Kim","year":"2023","journal-title":"Int J Med Inform"},{"key":"2026032708401080000_33.1.e101636.6","doi-asserted-by":"publisher","DOI":"10.1016\/j.ejmp.2021.04.016"},{"key":"2026032708401080000_33.1.e101636.7","doi-asserted-by":"crossref","first-page":"65138","DOI":"10.1109\/ACCESS.2023.3289839","article-title":"Pressure Ulcer Categorization and Reporting in Domiciliary Settings Using Deep Learning and Mobile Devices: A Clinical Trial to Evaluate End-to-End Performance","volume":"11","author":"Fergus","year":"2023","journal-title":"IEEE Access"},{"key":"2026032708401080000_33.1.e101636.8","doi-asserted-by":"crossref","first-page":"7124","DOI":"10.3390\/app14167124","article-title":"Diagnosis of Pressure Ulcer Stage Using On-Device AI","volume":"14","author":"Chang","year":"2024","journal-title":"Appl Sci (Basel)"},{"key":"2026032708401080000_33.1.e101636.9","doi-asserted-by":"crossref","DOI":"10.3389\/fmedt.2022.905074","article-title":"An artificial intelligence-enabled smartphone app for real-time pressure injury assessment","volume":"4","author":"Lau","year":"2022","journal-title":"Front Med Technol"},{"key":"2026032708401080000_33.1.e101636.10","unstructured":"Yonghye K . Yolo_Label (version1.2.1) [source code]. 2023. Available: https:\/\/github.com\/developer0hye\/Yolo_Label"},{"key":"2026032708401080000_33.1.e101636.11","unstructured":"European Pressure Ulcer Advisory Panel, National Pressure Injury Advisory Panel and Pan Pacific Pressure Injury Alliance . Prevention and treatment of pressure Ulcers\/injuries: clinical practice guideline. The international guideline. Emily haesler (ed.). EPUAP\/NPIAP\/PPPIA: 2019."},{"key":"2026032708401080000_33.1.e101636.12","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1186\/s40537-019-0197-0","article-title":"A survey on Image Data Augmentation for Deep Learning","volume":"6","author":"Shorten","year":"2019","journal-title":"J Big Data"},{"key":"2026032708401080000_33.1.e101636.13","doi-asserted-by":"crossref","unstructured":"Wang CY , Yeh IH , Mark Liao HY . Yolov9: learning what you want to learn using programmable gradient information. European conference on computer vision, Cham: Springer Nature Switzerland; 2024:1\u201321.","DOI":"10.1007\/978-3-031-72751-1_1"},{"key":"2026032708401080000_33.1.e101636.14","doi-asserted-by":"crossref","unstructured":"Huang G , Liu Z , Van Der Maaten L , et al . Densely connected convolutional networks. Proceedings of the IEEE conference on computer vision and pattern recognition; 2017:4700\u20138.","DOI":"10.1109\/CVPR.2017.243"},{"key":"2026032708401080000_33.1.e101636.15","unstructured":"Morgan N , Bourlard H . Generalization and parameter estimation in feedforward nets: some experiments. In: Advances in neural information processing systems, 2. 1989."},{"key":"2026032708401080000_33.1.e101636.16","unstructured":"Loshchilov I . Decoupled weight decay regularization. arXiv 2017. doi:10.48550\/arXiv.1711.05101"},{"key":"2026032708401080000_33.1.e101636.17","doi-asserted-by":"crossref","first-page":"1222","DOI":"10.3390\/healthcare11091222","article-title":"YOLO-Based Deep Learning Model for Pressure Ulcer Detection and Classification","volume":"11","author":"Aldughayfiq","year":"2023","journal-title":"Healthcare (Basel)"},{"key":"2026032708401080000_33.1.e101636.18","doi-asserted-by":"publisher","DOI":"10.1007\/s10278-014-9721-0"},{"key":"2026032708401080000_33.1.e101636.19","first-page":"415","article-title":"Constructing Inpatient Pressure Injury Prediction Models Using Machine Learning Techniques","volume":"38","author":"Hu","year":"2020","journal-title":"CIN"},{"key":"2026032708401080000_33.1.e101636.20","doi-asserted-by":"crossref","first-page":"16","DOI":"10.1097\/NNR.0000000000000258","article-title":"Hospital-Acquired Pressure Injury: Risk-Adjusted Comparisons in an Integrated Healthcare Delivery System","volume":"67","author":"Rondinelli","year":"2018","journal-title":"Nurs Res"}],"container-title":["BMJ Health &amp; Care Informatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.1136\/bmjhci-2025-101636","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,27]],"date-time":"2026-03-27T15:40:23Z","timestamp":1774626023000},"score":1,"resource":{"primary":{"URL":"https:\/\/informatics.bmj.com\/lookup\/doi\/10.1136\/bmjhci-2025-101636"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3]]},"references-count":20,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,3,27]]},"published-print":{"date-parts":[[2026,3]]}},"alternative-id":["10.1136\/bmjhci-2025-101636"],"URL":"https:\/\/doi.org\/10.1136\/bmjhci-2025-101636","relation":{},"ISSN":["2632-1009"],"issn-type":[{"value":"2632-1009","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,3]]}}}