Pub Date : 2024-08-26DOI: 10.1016/j.mcpdig.2024.08.007
Carlo Bulletti MD , Jason M. Franasiak MD , Andrea Busnelli MD , Romualdo Sciorio BSc, Msc , Marco Berrettini PhD , Lusine Aghajanova MD, PhD , Francesco M. Bulletti MD , Baris Ata MD
The aim of this systematic review was to identify clinical decision support algorithms (CDSAs) proposed for assisted reproductive technologies (ARTs) and to evaluate their effectiveness in improving ART cycles at every stage vs traditional methods, thereby providing an evidence-based guidance for their use in ART practice. A literature search on PubMed and Embase of articles published between 1 January 2013 and 31 January 2024 was performed to identify relevant articles. Prospective and retrospective studies in English on the use of CDSA for ART were included. Out of 1746 articles screened, 116 met the inclusion criteria. The selected articles were categorized into 3 areas: prognosis and patient counseling, clinical management, and embryo assessment. After screening, 11 CDSAs were identified as potentially valuable for clinical management and laboratory practices. Our findings highlight the potential of automated decision aids to improve in vitro fertilization outcomes. However, the main limitation of this review was the lack of standardization in validation methods across studies. Further validation and clinical trials are needed to establish the effectiveness of these tools in the clinical setting.
{"title":"Artificial Intelligence, Clinical Decision Support Algorithms, Mathematical Models, Calculators Applications in Infertility: Systematic Review and Hands-On Digital Applications","authors":"Carlo Bulletti MD , Jason M. Franasiak MD , Andrea Busnelli MD , Romualdo Sciorio BSc, Msc , Marco Berrettini PhD , Lusine Aghajanova MD, PhD , Francesco M. Bulletti MD , Baris Ata MD","doi":"10.1016/j.mcpdig.2024.08.007","DOIUrl":"10.1016/j.mcpdig.2024.08.007","url":null,"abstract":"<div><p>The aim of this systematic review was to identify clinical decision support algorithms (CDSAs) proposed for assisted reproductive technologies (ARTs) and to evaluate their effectiveness in improving ART cycles at every stage vs traditional methods, thereby providing an evidence-based guidance for their use in ART practice. A literature search on PubMed and Embase of articles published between 1 January 2013 and 31 January 2024 was performed to identify relevant articles. Prospective and retrospective studies in English on the use of CDSA for ART were included. Out of 1746 articles screened, 116 met the inclusion criteria. The selected articles were categorized into 3 areas: prognosis and patient counseling, clinical management, and embryo assessment. After screening, 11 CDSAs were identified as potentially valuable for clinical management and laboratory practices. Our findings highlight the potential of automated decision aids to improve in vitro fertilization outcomes. However, the main limitation of this review was the lack of standardization in validation methods across studies. Further validation and clinical trials are needed to establish the effectiveness of these tools in the clinical setting.</p></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"2 4","pages":"Pages 518-532"},"PeriodicalIF":0.0,"publicationDate":"2024-08-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2949761224000828/pdfft?md5=73ab3e5ee533ad8ca0f92b4e71a1331e&pid=1-s2.0-S2949761224000828-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142270756","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-26DOI: 10.1016/j.mcpdig.2024.08.005
Oana M. Dumitrascu MD, MSc , Xin Li MS , Wenhui Zhu MS , Bryan K. Woodruff MD , Simona Nikolova PhD , Jacob Sobczak , Amal Youssef MD , Siddhant Saxena , Janine Andreev , Richard J. Caselli MD , John J. Chen MD, PhD , Yalin Wang PhD
Objective
To report the development and performance of 2 distinct deep learning models trained exclusively on retinal color fundus photographs to classify Alzheimer disease (AD).
Patients and Methods
Two independent datasets (UK Biobank and our tertiary academic institution) of good-quality retinal photographs derived from patients with AD and controls were used to build 2 deep learning models, between April 1, 2021, and January 30, 2024. ADVAS is a U-Net–based architecture that uses retinal vessel segmentation. ADRET is a bidirectional encoder representations from transformers style self-supervised learning convolutional neural network pretrained on a large data set of retinal color photographs from UK Biobank. The models’ performance to distinguish AD from non-AD was determined using mean accuracy, sensitivity, specificity, and receiving operating curves. The generated attention heatmaps were analyzed for distinctive features.
Results
The self-supervised ADRET model had superior accuracy when compared with ADVAS, in both UK Biobank (98.27% vs 77.20%; P<.001) and our institutional testing data sets (98.90% vs 94.17%; P=.04). No major differences were noted between the original and binary vessel segmentation and between both eyes vs single-eye models. Attention heatmaps obtained from patients with AD highlighted regions surrounding small vascular branches as areas of highest relevance to the model decision making.
Conclusion
A bidirectional encoder representations from transformers style self-supervised convolutional neural network pretrained on a large data set of retinal color photographs alone can screen symptomatic AD with high accuracy, better than U-Net–pretrained models. To be translated in clinical practice, this methodology requires further validation in larger and diverse populations and integrated techniques to harmonize fundus photographs and attenuate the imaging-associated noise.
{"title":"Color Fundus Photography and Deep Learning Applications in Alzheimer Disease","authors":"Oana M. Dumitrascu MD, MSc , Xin Li MS , Wenhui Zhu MS , Bryan K. Woodruff MD , Simona Nikolova PhD , Jacob Sobczak , Amal Youssef MD , Siddhant Saxena , Janine Andreev , Richard J. Caselli MD , John J. Chen MD, PhD , Yalin Wang PhD","doi":"10.1016/j.mcpdig.2024.08.005","DOIUrl":"10.1016/j.mcpdig.2024.08.005","url":null,"abstract":"<div><h3>Objective</h3><p>To report the development and performance of 2 distinct deep learning models trained exclusively on retinal color fundus photographs to classify Alzheimer disease (AD).</p></div><div><h3>Patients and Methods</h3><p>Two independent datasets (UK Biobank and our tertiary academic institution) of good-quality retinal photographs derived from patients with AD and controls were used to build 2 deep learning models, between April 1, 2021, and January 30, 2024. ADVAS is a U-Net–based architecture that uses retinal vessel segmentation. ADRET is a bidirectional encoder representations from transformers style self-supervised learning convolutional neural network pretrained on a large data set of retinal color photographs from UK Biobank. The models’ performance to distinguish AD from non-AD was determined using mean accuracy, sensitivity, specificity, and receiving operating curves. The generated attention heatmaps were analyzed for distinctive features.</p></div><div><h3>Results</h3><p>The self-supervised ADRET model had superior accuracy when compared with ADVAS, in both UK Biobank (98.27% vs 77.20%; <em>P</em><.001) and our institutional testing data sets (98.90% vs 94.17%; <em>P</em>=.04). No major differences were noted between the original and binary vessel segmentation and between both eyes vs single-eye models. Attention heatmaps obtained from patients with AD highlighted regions surrounding small vascular branches as areas of highest relevance to the model decision making.</p></div><div><h3>Conclusion</h3><p>A bidirectional encoder representations from transformers style self-supervised convolutional neural network pretrained on a large data set of retinal color photographs alone can screen symptomatic AD with high accuracy, better than U-Net–pretrained models. To be translated in clinical practice, this methodology requires further validation in larger and diverse populations and integrated techniques to harmonize fundus photographs and attenuate the imaging-associated noise.</p></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"2 4","pages":"Pages 548-558"},"PeriodicalIF":0.0,"publicationDate":"2024-08-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2949761224000804/pdfft?md5=727bbdca5e1469575c30a5949adff677&pid=1-s2.0-S2949761224000804-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142270887","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-24DOI: 10.1016/j.mcpdig.2024.08.006
Ehab Harahsheh MBBS , Stephen W. English Jr. MD, MBA , Bart M. Demaerschalk MD , Kevin M. Barrett MD , William D. Freeman MD
Objective
To assess the feasibility and potential scalability of telemedicine-enabled ambulances for the prehospital evaluation of patients with suspected acute stroke symptoms.
Patients and Methods
A pilot study of telemedicine-enabled ambulances for evaluating patients with suspected acute stroke symptoms en route at 2 tertiary academic comprehensive stroke centers from January 1, 2018, to February 5, 2024. Charts of included patients were reviewed for demographic data, vascular risk factors, final diagnosis, time from arrival to neuroimaging, door-to–needle and door-to-puncture times in patients eligible for acute treatment, and any reported technical challenges during audio-video consultations.
Results
Forty-seven patients (mean age, 68.0 years; 62% men) were evaluated via telemedicine-enabled ambulances, of which 35 (74%) where for hospital-to-hospital transferred patients. Mean time from arrival to neuroimaging was 11.8 minutes. Twenty-nine patients (62%) were diagnosed with acute ischemic stroke, and the remainder were diagnosed with intracranial hemorrhage (n=13), seizure (n=2), brain mass (n=1), or other diagnoses (n=3). Four patients (9%) received intravenous thrombolysis with alteplase (mean door to needle, 30.3 minutes), and 15 patients (32%) underwent mechanical thrombectomy (mean door to puncture, 72 minutes). Technical challenges were reported in 15 of the 42 (36%) cases, of which 10 (67%) were related to poor internet connectivity.
Conclusion
Telemedicine-enabled ambulances in emergency medical services systems are novel, feasible, and potentially scalable options for evaluating patients with suspected acute stroke in the prehospital setting. However, optimization of infrastructure, staffing models, and internet connectivity is necessary, and larger studies evaluating the clinical and cost effectiveness of this approach are needed before widespread implementation.
{"title":"Telemedicine-Enabled Ambulances for Prehospital Acute Stroke Management: A Pilot Study","authors":"Ehab Harahsheh MBBS , Stephen W. English Jr. MD, MBA , Bart M. Demaerschalk MD , Kevin M. Barrett MD , William D. Freeman MD","doi":"10.1016/j.mcpdig.2024.08.006","DOIUrl":"10.1016/j.mcpdig.2024.08.006","url":null,"abstract":"<div><h3>Objective</h3><p>To assess the feasibility and potential scalability of telemedicine-enabled ambulances for the prehospital evaluation of patients with suspected acute stroke symptoms.</p></div><div><h3>Patients and Methods</h3><p>A pilot study of telemedicine-enabled ambulances for evaluating patients with suspected acute stroke symptoms en route at 2 tertiary academic comprehensive stroke centers from January 1, 2018, to February 5, 2024. Charts of included patients were reviewed for demographic data, vascular risk factors, final diagnosis, time from arrival to neuroimaging, door-to–needle and door-to-puncture times in patients eligible for acute treatment, and any reported technical challenges during audio-video consultations.</p></div><div><h3>Results</h3><p>Forty-seven patients (mean age, 68.0 years; 62% men) were evaluated via telemedicine-enabled ambulances, of which 35 (74%) where for hospital-to-hospital transferred patients. Mean time from arrival to neuroimaging was 11.8 minutes. Twenty-nine patients (62%) were diagnosed with acute ischemic stroke, and the remainder were diagnosed with intracranial hemorrhage (n=13), seizure (n=2), brain mass (n=1), or other diagnoses (n=3). Four patients (9%) received intravenous thrombolysis with alteplase (mean door to needle, 30.3 minutes), and 15 patients (32%) underwent mechanical thrombectomy (mean door to puncture, 72 minutes). Technical challenges were reported in 15 of the 42 (36%) cases, of which 10 (67%) were related to poor internet connectivity.</p></div><div><h3>Conclusion</h3><p>Telemedicine-enabled ambulances in emergency medical services systems are novel, feasible, and potentially scalable options for evaluating patients with suspected acute stroke in the prehospital setting. However, optimization of infrastructure, staffing models, and internet connectivity is necessary, and larger studies evaluating the clinical and cost effectiveness of this approach are needed before widespread implementation.</p></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"2 4","pages":"Pages 533-541"},"PeriodicalIF":0.0,"publicationDate":"2024-08-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2949761224000816/pdfft?md5=64c6e2a698aeff6e85f5c59a0753402b&pid=1-s2.0-S2949761224000816-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142270885","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-17DOI: 10.1016/j.mcpdig.2024.07.004
Shankargouda Patil BDS, MDS, PhD, Frank W. Licari MBA, DDS, MPH
{"title":"Can Artificial Intelligence Make the Cut? Dissecting Large Language Model’s Surgical Exam Performance","authors":"Shankargouda Patil BDS, MDS, PhD, Frank W. Licari MBA, DDS, MPH","doi":"10.1016/j.mcpdig.2024.07.004","DOIUrl":"10.1016/j.mcpdig.2024.07.004","url":null,"abstract":"","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"2 4","pages":"Page 515"},"PeriodicalIF":0.0,"publicationDate":"2024-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2949761224000774/pdfft?md5=bc46f5c781e888252e2a710c49f093ca&pid=1-s2.0-S2949761224000774-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142270757","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-13DOI: 10.1016/j.mcpdig.2024.08.003
Adam M. Ostrovsky BS, Joshua R. Chen BS, Vishal N. Shah DO, Babak Abai MD
{"title":"In Reply: Can Artificial Intelligence Make the Cut? Dissecting Large Language Model’s Surgical Exam Performance","authors":"Adam M. Ostrovsky BS, Joshua R. Chen BS, Vishal N. Shah DO, Babak Abai MD","doi":"10.1016/j.mcpdig.2024.08.003","DOIUrl":"10.1016/j.mcpdig.2024.08.003","url":null,"abstract":"","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"2 4","pages":"Pages 516-517"},"PeriodicalIF":0.0,"publicationDate":"2024-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2949761224000786/pdfft?md5=cf07bf2d7ac7814d8b6badf50c157ce5&pid=1-s2.0-S2949761224000786-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142270906","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-10DOI: 10.1016/j.mcpdig.2024.08.001
Jose K. James MD, PhD , Tharana Maran MS , Mindy P. Rice MBA , Tanner S Hunt MHA , Kevin J. Peterson PhD, MS , William J. Hogan MBBCh , Shivam Damani BS , Alexander J. Ryu MD
{"title":"Experience With an Optical Character Recognition Search Application for Review of Outside Medical Records","authors":"Jose K. James MD, PhD , Tharana Maran MS , Mindy P. Rice MBA , Tanner S Hunt MHA , Kevin J. Peterson PhD, MS , William J. Hogan MBBCh , Shivam Damani BS , Alexander J. Ryu MD","doi":"10.1016/j.mcpdig.2024.08.001","DOIUrl":"10.1016/j.mcpdig.2024.08.001","url":null,"abstract":"","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"2 4","pages":"Pages 511-514"},"PeriodicalIF":0.0,"publicationDate":"2024-08-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2949761224000750/pdfft?md5=f972be308efb8a3a09a756ace0999dab&pid=1-s2.0-S2949761224000750-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142270883","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-08DOI: 10.1016/j.mcpdig.2024.07.003
Levi W. Disrud , Tara A. Gosse MS , Zach D. Linn MS , Anthony H. Kashou MD , Peter A. Noseworthy MD, MBA , Angela Fink MSN , Dawn Griffin MA, MBA , Blade Faust
Objective
To investigate the operational outcomes and implementation effects of tiered cardiac telemetry monitoring in a hospital environment using an innovative technology.
Patients and Methods
The research focuses on assessing the precision, speed, and reliability of alerts generated by a wireless device in adult patients aged 18 and above, concurrently monitored by a hardwired, continuous cardiac monitor. Using an agile methodology, we tested and validated a nonhardwired, cellular-connected continuous cardiac monitor (InfoBionic MoMe) in 162 patients. A comparison was made between the wireless device and the standard hardwired system, conducted at Mayo Clinic Hospital with Institutional Review Board approval from June 6, 2022, to December 15, 2022.
Results
The study revealed a high correlation of events captured compared with the standard care model. Differences in algorithms, alarm parameters, and operational considerations impacting clinical implementation were observed. Connectivity improvements during the study reduced latency from 3-5 minutes to 30 seconds. Delayed alarms were attributed to device damage (4.5% of cases) and poor cellular connections (29% within 31-60 seconds).
Conclusion
The implementation of tiered cardiac telemetry in hospital environments, coupled with advancements in remote cardiac monitoring, supports expanded bedside telemetry capabilities and near real-time remote monitoring postdischarge. Although the study successfully validated the wireless device concept, improvements are needed before implementation for inpatient cardiac monitoring. Further research and technological enhancements can build on these findings to enhance health care practices in this domain.
{"title":"Implementation of a Tiered Cardiac Telemetry System: An Operational Blueprint at Mayo Clinic","authors":"Levi W. Disrud , Tara A. Gosse MS , Zach D. Linn MS , Anthony H. Kashou MD , Peter A. Noseworthy MD, MBA , Angela Fink MSN , Dawn Griffin MA, MBA , Blade Faust","doi":"10.1016/j.mcpdig.2024.07.003","DOIUrl":"10.1016/j.mcpdig.2024.07.003","url":null,"abstract":"<div><h3>Objective</h3><p>To investigate the operational outcomes and implementation effects of tiered cardiac telemetry monitoring in a hospital environment using an innovative technology.</p></div><div><h3>Patients and Methods</h3><p>The research focuses on assessing the precision, speed, and reliability of alerts generated by a wireless device in adult patients aged 18 and above, concurrently monitored by a hardwired, continuous cardiac monitor. Using an agile methodology, we tested and validated a nonhardwired, cellular-connected continuous cardiac monitor (InfoBionic MoMe) in 162 patients. A comparison was made between the wireless device and the standard hardwired system, conducted at Mayo Clinic Hospital with Institutional Review Board approval from June 6, 2022, to December 15, 2022.</p></div><div><h3>Results</h3><p>The study revealed a high correlation of events captured compared with the standard care model. Differences in algorithms, alarm parameters, and operational considerations impacting clinical implementation were observed. Connectivity improvements during the study reduced latency from 3-5 minutes to 30 seconds. Delayed alarms were attributed to device damage (4.5% of cases) and poor cellular connections (29% within 31-60 seconds).</p></div><div><h3>Conclusion</h3><p>The implementation of tiered cardiac telemetry in hospital environments, coupled with advancements in remote cardiac monitoring, supports expanded bedside telemetry capabilities and near real-time remote monitoring postdischarge. Although the study successfully validated the wireless device concept, improvements are needed before implementation for inpatient cardiac monitoring. Further research and technological enhancements can build on these findings to enhance health care practices in this domain.</p></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"2 4","pages":"Pages 542-547"},"PeriodicalIF":0.0,"publicationDate":"2024-08-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2949761224000749/pdfft?md5=c1d48c297b7210b8e9ed902ed5d8b9a6&pid=1-s2.0-S2949761224000749-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142270886","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-06DOI: 10.1016/j.mcpdig.2024.06.010
Mauricio F. Jin MD , Peter A. Noseworthy MD , Xiaoxi Yao PhD
The emergence of artificial intelligence (AI) and other digital solutions in health care has considerably altered the landscape of medical research and patient care. Rigorous evaluation in routine practice settings is fundamental to the ethical use of AI and consists of 3 stages of evaluations: technical performance, usability and acceptability, and health impact evaluation. Pragmatic trials often play a key role in the health impact evaluation. The current review introduces the concept of pragmatic trials, their role in AI evaluation, the challenges of conducting pragmatic trials, and strategies to mitigate the challenges. We also examined common designs used in pragmatic trials and highlighted examples of published or ongoing AI trials. As more health systems advance into learning health systems, where outcomes are continuously evaluated to refine processes and tools, pragmatic trials embedded into everyday practice, leveraging data and infrastructure from delivering health care, will be a critical part of the feedback cycle for learning and improvement.
{"title":"Assessing Artificial Intelligence Solution Effectiveness: The Role of Pragmatic Trials","authors":"Mauricio F. Jin MD , Peter A. Noseworthy MD , Xiaoxi Yao PhD","doi":"10.1016/j.mcpdig.2024.06.010","DOIUrl":"10.1016/j.mcpdig.2024.06.010","url":null,"abstract":"<div><p>The emergence of artificial intelligence (AI) and other digital solutions in health care has considerably altered the landscape of medical research and patient care. Rigorous evaluation in routine practice settings is fundamental to the ethical use of AI and consists of 3 stages of evaluations: technical performance, usability and acceptability, and health impact evaluation. Pragmatic trials often play a key role in the health impact evaluation. The current review introduces the concept of pragmatic trials, their role in AI evaluation, the challenges of conducting pragmatic trials, and strategies to mitigate the challenges. We also examined common designs used in pragmatic trials and highlighted examples of published or ongoing AI trials. As more health systems advance into learning health systems, where outcomes are continuously evaluated to refine processes and tools, pragmatic trials embedded into everyday practice, leveraging data and infrastructure from delivering health care, will be a critical part of the feedback cycle for learning and improvement.</p></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"2 4","pages":"Pages 499-510"},"PeriodicalIF":0.0,"publicationDate":"2024-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2949761224000737/pdfft?md5=af45d105d6bd843dd364187f67b18e58&pid=1-s2.0-S2949761224000737-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142270884","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-07-31DOI: 10.1016/j.mcpdig.2024.07.002
Yashbir Singh PhD , Shahriar Faghani MD , John E. Eaton MD , Sudhakar K. Venkatesh MD , Bradley J. Erickson MD, PhD
Objective
To investigate a deep learning model for predicting hepatic decompensation using computed tomography (CT) imaging in patients with primary sclerosing cholangitis (PSC).
Patients and Methods
Retrospective cohort study involving 277 adult patients with large-duct PSC who underwent an abdominal CT scan. The portal venous phase CT images were used as input to a 3D-DenseNet121 model, which was trained using 5-fold crossvalidation to classify hepatic decompensation. To further investigate the role of each anatomic region in the model’s decision-making process, we trained the model on different sections of 3-dimensional CT images. This included training on the right, left, anterior, posterior, inferior, and superior halves of the image data set. For each half, as well as for the entire scan, we performed area under the receiving operating curve (AUROC) analysis.
Results
Hepatic decompensation occurred in 128 individuals after a median (interquartile range) of 1.5 years (142-1318 days) after the CT scan. The deep learning model exhibited promising results, with a mean ± SD AUROC of 0.89±0.04 for the baseline model. The mean ± SD AUROC for left, right, anterior, posterior, superior, and inferior halves were 0.83±0.03, 0.83±0.03, 0.82±0.09, 0.79±0.02, 0.78±0.02, and 0.76±0.04, respectively.
Conclusion
The study illustrates the potential of examining CT imaging using 3D-DenseNet121 deep learning model to predict hepatic decompensation in patients with PSC.
{"title":"Deep Learning–Based Prediction of Hepatic Decompensation in Patients With Primary Sclerosing Cholangitis With Computed Tomography","authors":"Yashbir Singh PhD , Shahriar Faghani MD , John E. Eaton MD , Sudhakar K. Venkatesh MD , Bradley J. Erickson MD, PhD","doi":"10.1016/j.mcpdig.2024.07.002","DOIUrl":"10.1016/j.mcpdig.2024.07.002","url":null,"abstract":"<div><h3>Objective</h3><p>To investigate a deep learning model for predicting hepatic decompensation using computed tomography (CT) imaging in patients with primary sclerosing cholangitis (PSC).</p></div><div><h3>Patients and Methods</h3><p>Retrospective cohort study involving 277 adult patients with large-duct PSC who underwent an abdominal CT scan. The portal venous phase CT images were used as input to a 3D-DenseNet121 model, which was trained using 5-fold crossvalidation to classify hepatic decompensation. To further investigate the role of each anatomic region in the model’s decision-making process, we trained the model on different sections of 3-dimensional CT images. This included training on the right, left, anterior, posterior, inferior, and superior halves of the image data set. For each half, as well as for the entire scan, we performed area under the receiving operating curve (AUROC) analysis.</p></div><div><h3>Results</h3><p>Hepatic decompensation occurred in 128 individuals after a median (interquartile range) of 1.5 years (142-1318 days) after the CT scan. The deep learning model exhibited promising results, with a mean ± SD AUROC of 0.89±0.04 for the baseline model. The mean ± SD AUROC for left, right, anterior, posterior, superior, and inferior halves were 0.83±0.03, 0.83±0.03, 0.82±0.09, 0.79±0.02, 0.78±0.02, and 0.76±0.04, respectively.</p></div><div><h3>Conclusion</h3><p>The study illustrates the potential of examining CT imaging using 3D-DenseNet121 deep learning model to predict hepatic decompensation in patients with PSC.</p></div>","PeriodicalId":74127,"journal":{"name":"Mayo Clinic Proceedings. Digital health","volume":"2 3","pages":"Pages 470-476"},"PeriodicalIF":0.0,"publicationDate":"2024-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2949761224000725/pdfft?md5=b6cb80150bd80f9c0ac6702b4e71c527&pid=1-s2.0-S2949761224000725-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142087151","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}