{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,15]],"date-time":"2026-10-15T20:06:28Z","timestamp":1792094788186,"version":"4.4.0"},"reference-count":34,"publisher":"JMIR Publications Inc.","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["JMIR Med Inform"],"abstract":"Abstract<\/jats:title>\n \n Background<\/jats:title>\n Visual identification and verification of medications during dispensing and administration are prone to human error, particularly in high-pressure and high-volume clinical settings. Misidentification can lead to medication errors, posing risks to patient safety and placing a burden on health care systems. Recent advances in computer vision and object detection offer promising solutions for automated solid oral dosage form (pill) recognition. However, comprehensive studies comparing code-based and no-code (automated machine learning [AutoML]) approaches for pill recognition are lacking.<\/jats:p>\n <\/jats:sec>\n \n Objective<\/jats:title>\n This study aimed to evaluate and compare the performance, cost, usability, and deployment feasibility of pill recognition models developed with Ultralytics YOLO11 and 3 cloud-based AutoML platforms (Amazon Rekognition Custom Labels, Google Vertex artificial intelligence [AI] AutoML Vision, and Microsoft Azure Custom Vision) using multiple datasets, including real-world clinical images.<\/jats:p>\n <\/jats:sec>\n \n Methods<\/jats:title>\n Five training subsets of increasing size (1230, 3450, 7380, 14,400, and 26,880 images) from 30 commonly dispensed medications were used to train models on YOLO11 and 3 AutoML platforms. Models were evaluated on 6 datasets from different environments: clinical images from 3 hospitals, a verification dataset, a laboratory dataset, and an exhaustive testing set. Performance metrics, including accuracy, precision, recall, and mean average precision, were calculated. We evaluated the impact of training data size on performance and benchmarked training time, platform costs, and limitations.<\/jats:p>\n <\/jats:sec>\n \n Results<\/jats:title>\n No single platform dominated across all test environments. On the verification dataset (optimal conditions), accuracy ranged from 80.83% (YOLO11) to 91.60% (Google Vertex AI) when trained with the full training dataset. YOLO11 showed consistent performance improvement with increasing training data (accuracy: 63.06%-80.83%) and achieved near-perfect precision and mean average precision scores (0.95\u20101.00). Google Vertex AI reached above 90% accuracy on 3 training subsets but showed unpredictable declines. Amazon Rekognition maintained near-perfect precision (0.92\u20101.00) but had the highest false negative rates (up to 0.74), missing many pills. Custom Vision demonstrated steady performance improvements (77.08%-85.62% accuracy) but lagged behind other AutoML platforms, probably due to its older YOLOv2-based architecture. On clinical datasets, accuracy fluctuated (20.62%-90%) depending on the dataset and platform. Training costs and time varied: YOLO11 (open-source), Microsoft Azure (US $9.50-US $28.60, allowed user-predefined training duration), Google Vertex AI (US $69.30 with consistent 2.5\u20103-hour training times), and Amazon Rekognition (US $5.43-US $43.89 with size-dependent training time scaling, reaching nearly 40 hours on the full 26,880-image dataset).<\/jats:p>\n <\/jats:sec>\n \n Conclusions<\/jats:title>\n Each platform offers distinct advantages and trade-offs: YOLO11 provides the highest flexibility and lowest platform costs but requires technical expertise, while AutoML platforms can offer high performance at a higher cost but with limited user control, introducing unpredictability. The performance variations demonstrate that successful clinical deployment requires careful platform selection based on specific performance requirements, budget constraints, and available technical resources, followed by rigorous validation using real-world, representative data to ensure patient safety in clinical workflows.<\/jats:p>\n <\/jats:sec>","DOI":"10.2196\/79160","type":"journal-article","created":{"date-parts":[[2026,2,17]],"date-time":"2026-02-17T15:30:07Z","timestamp":1771342207000},"page":"e79160-e79160","source":"Crossref","is-referenced-by-count":2,"title":["Code-Based Versus AutoML Methods for Pill Recognition in Clinical Settings: Comparative Performance Study"],"prefix":"10.2196","volume":"14","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-0339-5653","authenticated-orcid":false,"given":"Amir Reza","family":"Ashraf","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0009-3160-1275","authenticated-orcid":false,"given":"Rich\u00e1rd","family":"R\u00e1dli","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0004-3032-8784","authenticated-orcid":false,"given":"Zsolt","family":"V\u00f6r\u00f6sh\u00e1zi","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8243-6179","authenticated-orcid":false,"given":"Andr\u00e1s","family":"Fittler","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1010","published-online":{"date-parts":[[2026,4,10]]},"reference":[{"key":"R1","unstructured":"Global burden of preventable medication-related harm in health care: a systematic review. World Health Organization. 2023. URL: https:\/\/www.who.int\/publications\/i\/item\/9789240088887 [Accessed 16-06-2025]"},{"issue":"2","key":"R2","doi-asserted-by":"publisher","first-page":"96","DOI":"10.1136\/bmjqs-2019-010206","article-title":"Economic analysis of the prevalence and clinical and economic burden of medication error in England","volume":"30","author":"Elliott","journal-title":"BMJ Qual Saf"},{"issue":"9","key":"R3","doi-asserted-by":"publisher","first-page":"1183","DOI":"10.1093\/jamia\/ocy066","article-title":"The national cost of adverse drug events resulting from inappropriate medication-related alert overrides in the United States","volume":"25","author":"Slight","journal-title":"J Am Med Inform Assoc"},{"issue":"1","key":"R4","doi-asserted-by":"publisher","DOI":"10.1186\/s12916-020-01774-9","article-title":"Preventable medication harm across health care settings: a systematic review and meta-analysis","volume":"18","author":"Hodkinson","journal-title":"BMC Med"},{"issue":"3","key":"R5","doi-asserted-by":"publisher","first-page":"322","DOI":"10.2174\/1574886315666201002124713","article-title":"Role and services of a pharmacist in the prevention of medication errors: a systematic review","volume":"16","author":"Gillani","journal-title":"Curr Drug Saf"},{"issue":"1","key":"R6","doi-asserted-by":"publisher","first-page":"118","DOI":"10.1038\/s41746-021-00483-8","article-title":"Performance evaluation of a prescription medication image classification model: an observational cohort","volume":"4","author":"Lester","journal-title":"NPJ Digit Med"},{"key":"R7","doi-asserted-by":"publisher","DOI":"10.1016\/j.apergo.2022.103831","article-title":"Improving safety in the operating room: medication icon labels increase visibility and discrimination","volume":"104","author":"Lusk","journal-title":"Appl Ergon"},{"issue":"2","key":"R8","doi-asserted-by":"publisher","first-page":"58","DOI":"10.1136\/ejhpharm-2019-002081","article-title":"Automation of in-hospital pharmacy dispensing: a systematic review","volume":"28","author":"Batson","journal-title":"Eur J Hosp Pharm"},{"issue":"3","key":"R9","doi-asserted-by":"publisher","first-page":"127","DOI":"10.1136\/ejhpharm-2021-003002","article-title":"Automated unit dose dispensing systems producing individually packaged and labelled drugs for inpatients: a systematic review","volume":"30","author":"H\u00e4nninen","journal-title":"Eur J Hosp Pharm"},{"issue":"6","key":"R10","doi-asserted-by":"publisher","first-page":"9243","DOI":"10.1007\/s11042-022-13644-y","article-title":"Object detection using YOLO: challenges, architectural successors, datasets and applications","volume":"82","author":"Diwan","journal-title":"Multimed Tools Appl"},{"key":"R11","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2024.3386826","article-title":"A comprehensive systematic review of YOLO for medical object detection (2018 to 2023)","volume":"12","author":"Ragab","journal-title":"IEEE Access"},{"issue":"1","key":"R12","doi-asserted-by":"publisher","DOI":"10.1186\/s12911-021-01691-8","article-title":"Comparison of RetinaNet, SSD, and YOLO v3 for real-time pill identification","volume":"21","author":"Tan","journal-title":"BMC Med Inform Decis Mak"},{"issue":"14","key":"R13","doi-asserted-by":"publisher","first-page":"4623","DOI":"10.1049\/ipr2.13273","article-title":"Metric\u2010based pill recognition with the help of textual and visual cues","volume":"18","author":"R\u00e1dli","journal-title":"IET Image Process"},{"key":"R14","unstructured":"Khanam R Hussain M . YOLOv11: an overview of the key architectural enhancements. arXiv. Preprint posted online on Oct 23, 2024. URL: https:\/\/arxiv.org\/abs\/2410.17725 [Accessed 14-04-2025] [doi: 10.48550\/arXiv.2410.17725]"},{"key":"R15","doi-asserted-by":"publisher","DOI":"10.1016\/j.artmed.2024.102844","article-title":"Leveraging code-free deep learning for pill recognition in clinical settings: a multicenter, real-world study of performance across multiple platforms","volume":"150","author":"Ashraf","journal-title":"Artif Intell Med"},{"key":"R16","doi-asserted-by":"publisher","DOI":"10.1016\/j.imu.2021.100607","article-title":"An easy method for identifying 315 categories of commonly-used Chinese herbal medicines based on automated image recognition using AutoML platforms","volume":"25","author":"Chen","journal-title":"Inform Med Unlocked"},{"issue":"4","key":"R17","doi-asserted-by":"publisher","first-page":"1381","DOI":"10.21037\/qims-20-922","article-title":"Evaluation of the performance of traditional machine learning algorithms, convolutional neural network and AutoML vision in ultrasound breast lesions classification: a comparative study","volume":"11","author":"Wan","journal-title":"Quant Imaging Med Surg"},{"key":"R18","unstructured":"Sapkota R Qureshi R Flores-Calero M et al. YOLO11 to its genesis: a decadal and comprehensive review of the You Only Look Once (YOLO) series. arXiv. Preprint posted online on Jun 12, 2024. URL: https:\/\/arxiv.org\/abs\/2406.19407v5 [Accessed 16-04-2025] [doi: 10.48550\/arXiv.2406.19407]"},{"issue":"5","key":"R19","first-page":"107","author":"Ayyalasomayajula","journal-title":"Int J Comput Sci Trends Technol"},{"key":"R20","doi-asserted-by":"crossref","unstructured":"Sun Y Song Q Gui X Ma F Wang T . AutoML in the wild: obstacles, workarounds, and expectations. Presented at: Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems; Apr 23-28, 2023:1-15; Hamburg, Germany. Apr 19, 2023.[doi: 10.1145\/3544548.3581082]","DOI":"10.1145\/3544548.3581082"},{"key":"R21","unstructured":"Microsoft. What is Custom Vision?. URL: https:\/\/learn.microsoft.com\/en-us\/azure\/ai-services\/custom-vision-service\/overview [Accessed 17-04-2025]"},{"key":"R22","unstructured":"Vertex AI documentation. Google. URL: https:\/\/cloud.google.com\/vertex-ai\/docs [Accessed 24-04-2025]"},{"key":"R23","unstructured":"Interfaces for Vertex AI. Google. URL: https:\/\/cloud.google.com\/vertex-ai\/docs\/start\/introduction-interfaces [Accessed 24-04-2025]"},{"key":"R24","unstructured":"Amazon Web Services. What is Amazon Rekognition Custom Labels? URL: https:\/\/docs.aws.amazon.com\/rekognition\/latest\/customlabels-dg\/what-is.html [Accessed 24-04-2025]"},{"key":"R25","doi-asserted-by":"crossref","unstructured":"Ling S Pastor A Li J et al. Few-shot pill recognition. Presented at: 2020 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR); Jun 13-19, 2020; Seattle, WA. [doi: 10.1109\/CVPR42600.2020.00981]","DOI":"10.1109\/CVPR42600.2020.00981"},{"key":"R26","doi-asserted-by":"publisher","DOI":"10.1109\/AIPR.2016.8010584","article-title":"The National Library of Medicine pill image recognition challenge: an initial report","volume":"2016","author":"Yaniv","journal-title":"IEEE Appl Imag Pattern Recognit Workshop"},{"key":"R27","unstructured":"Microsoft. What\u2019s new in Custom Vision?. URL: https:\/\/learn.microsoft.com\/en-us\/azure\/ai-services\/custom-vision-service\/whats-new [Accessed 17-04-2025]"},{"key":"R28","unstructured":"Prepare image training data for object detection. Google. URL: https:\/\/cloud.google.com\/vertex-ai\/docs\/image-data\/object-detection\/prepare-data [Accessed 24-04-2025]"},{"key":"R29","unstructured":"About data splits for AutoML models. Google. URL: https:\/\/cloud.google.com\/vertex-ai\/docs\/general\/ml-use [Accessed 24-04-2025]"},{"key":"R30","unstructured":"Managing labels. Amazon Web Services. URL: https:\/\/docs.aws.amazon.com\/rekognition\/latest\/customlabels-dg\/md-labels.html [Accessed 24-04-2025]"},{"key":"R31","unstructured":"Creating training and test datasets with images. Amazon Web Services. URL: https:\/\/docs.aws.amazon.com\/rekognition\/latest\/customlabels-dg\/md-create-dataset.html [Accessed 24-04-2025]"},{"key":"R32","unstructured":"Pricing. Google. URL: https:\/\/cloud.google.com\/vertex-ai\/pricing [Accessed 14-06-2025]"},{"key":"R33","unstructured":"Amazon Web Services. Amazon Rekognition pricing. URL: https:\/\/aws.amazon.com\/rekognition\/pricing\/ [Accessed 14-06-2025]"},{"key":"R34","unstructured":"Azure AI Custom Vision Service. Microsoft. URL: https:\/\/azure.microsoft.com\/en-us\/pricing\/details\/cognitive-services\/custom-vision-service\/ [Accessed 14-06-2025]"}],"container-title":["JMIR Medical Informatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/medinform.jmir.org\/2026\/1\/e79160","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,30]],"date-time":"2026-06-30T18:12:03Z","timestamp":1782843123000},"score":1,"resource":{"primary":{"URL":"https:\/\/medinform.jmir.org\/2026\/1\/e79160"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,10]]},"references-count":34,"URL":"https:\/\/doi.org\/10.2196\/79160","relation":{},"ISSN":["2291-9694"],"issn-type":[{"value":"2291-9694","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,4,10]]},"article-number":"v14i2e79160"}}