Saki Ito, Michal Cohen-Shelly, Zachi I Attia, Eunjung Lee, Paul A Friedman, Vuyisile T Nkomo, Hector I Michelena, Peter A Noseworthy, Francisco Lopez-Jimenez, Jae K Oh
Aims: An artificial intelligence-enabled electrocardiogram (AI-ECG) is a promising tool to detect patients with aortic stenosis (AS) before developing symptoms. However, functional, structural, or haemodynamic components reflected in AI-ECG responsible for its detection are unknown.
Methods and results: The AI-ECG model that was developed at Mayo Clinic using a convolutional neural network to identify patients with moderate-severe AS was applied. In patients used as the testing group, the correlation between the AI-ECG probability of AS and echocardiographic parameters was investigated. This study included 102 926 patients (63.0 ± 16.3 years, 52% male), and 28 464 (27.7%) were identified as AS positive by AI-ECG. Older age, atrial fibrillation, hypertension, diabetes, coronary artery disease, and heart failure were more common in the positive AI-ECG group than in the negative group (P < 0.001). The AI-ECG was correlated with aortic valve area (ρ = -0.48, R2 = 0.20), peak velocity (ρ = 0.22, R2 = 0.08), and mean pressure gradient (ρ = 0.35, R2 = 0.08). The AI-ECG also correlated with left ventricular (LV) mass index (ρ = 0.36, R2 = 0.13), E/e' (ρ = 0.36, R2 = 0.12), and left atrium volume index (ρ = 0.42, R2 = 0.12). Neither LV ejection fraction nor stroke volume index had a significant correlation with the AI-ECG. Age correlated with the AI-ECG (ρ = 0.46, R2 = 0.22) and its correlation with echocardiography parameters was similar to that of the AI-ECG.
Conclusion: A combination of AS severity, diastolic dysfunction, and LV hypertrophy is reflected in the AI-ECG to detect AS. There seems to be a gradation of the cardiac anatomical/functional features in the model and its identification process of AS is multifactorial.
{"title":"Correlation between artificial intelligence-enabled electrocardiogram and echocardiographic features in aortic stenosis.","authors":"Saki Ito, Michal Cohen-Shelly, Zachi I Attia, Eunjung Lee, Paul A Friedman, Vuyisile T Nkomo, Hector I Michelena, Peter A Noseworthy, Francisco Lopez-Jimenez, Jae K Oh","doi":"10.1093/ehjdh/ztad009","DOIUrl":"https://doi.org/10.1093/ehjdh/ztad009","url":null,"abstract":"<p><strong>Aims: </strong>An artificial intelligence-enabled electrocardiogram (AI-ECG) is a promising tool to detect patients with aortic stenosis (AS) before developing symptoms. However, functional, structural, or haemodynamic components reflected in AI-ECG responsible for its detection are unknown.</p><p><strong>Methods and results: </strong>The AI-ECG model that was developed at Mayo Clinic using a convolutional neural network to identify patients with moderate-severe AS was applied. In patients used as the testing group, the correlation between the AI-ECG probability of AS and echocardiographic parameters was investigated. This study included 102 926 patients (63.0 ± 16.3 years, 52% male), and 28 464 (27.7%) were identified as AS positive by AI-ECG. Older age, atrial fibrillation, hypertension, diabetes, coronary artery disease, and heart failure were more common in the positive AI-ECG group than in the negative group (<i>P</i> < 0.001). The AI-ECG was correlated with aortic valve area (ρ = -0.48, <i>R</i><sup>2</sup> = 0.20), peak velocity (ρ = 0.22, <i>R</i><sup>2</sup> = 0.08), and mean pressure gradient (ρ = 0.35, <i>R</i><sup>2</sup> = 0.08). The AI-ECG also correlated with left ventricular (LV) mass index (ρ = 0.36, <i>R</i><sup>2</sup> = 0.13), <i>E</i>/<i>e</i>' (ρ = 0.36, <i>R</i><sup>2</sup> = 0.12), and left atrium volume index (ρ = 0.42, <i>R</i><sup>2</sup> = 0.12). Neither LV ejection fraction nor stroke volume index had a significant correlation with the AI-ECG. Age correlated with the AI-ECG (ρ = 0.46, <i>R</i><sup>2</sup> = 0.22) and its correlation with echocardiography parameters was similar to that of the AI-ECG.</p><p><strong>Conclusion: </strong>A combination of AS severity, diastolic dysfunction, and LV hypertrophy is reflected in the AI-ECG to detect AS. There seems to be a gradation of the cardiac anatomical/functional features in the model and its identification process of AS is multifactorial.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ftp.ncbi.nlm.nih.gov/pub/pmc/oa_pdf/09/07/ztad009.PMC10232245.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"9571917","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}
This editorial refers to ‘A novel breakthrough in wrist-worn transdermal troponin-I-sensor assessment for acute myocardial infarction’
{"title":"Infrared spectral analysis for the classification of patients with acute coronary syndrome. The questions run so deep.","authors":"Alessandra Scoccia, Peter de Jaegere","doi":"10.1093/ehjdh/ztad017","DOIUrl":"https://doi.org/10.1093/ehjdh/ztad017","url":null,"abstract":"This editorial refers to ‘A novel breakthrough in wrist-worn transdermal troponin-I-sensor assessment for acute myocardial infarction’","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10232277/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"9571919","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}
Tobias Paul Seraphin, Mark Luedde, Christoph Roderburg, Marko van Treeck, Pascal Scheider, Roman D Buelow, Peter Boor, Sven H Loosen, Zdenek Provaznik, Daniel Mendelsohn, Filip Berisha, Christina Magnussen, Dirk Westermann, Tom Luedde, Christoph Brochhausen, Samuel Sossalla, Jakob Nikolas Kather
Aims: One of the most important complications of heart transplantation is organ rejection, which is diagnosed on endomyocardial biopsies by pathologists. Computer-based systems could assist in the diagnostic process and potentially improve reproducibility. Here, we evaluated the feasibility of using deep learning in predicting the degree of cellular rejection from pathology slides as defined by the International Society for Heart and Lung Transplantation (ISHLT) grading system.
Methods and results: We collected 1079 histopathology slides from 325 patients from three transplant centres in Germany. We trained an attention-based deep neural network to predict rejection in the primary cohort and evaluated its performance using cross-validation and by deploying it to three cohorts. For binary prediction (rejection yes/no), the mean area under the receiver operating curve (AUROC) was 0.849 in the cross-validated experiment and 0.734, 0.729, and 0.716 in external validation cohorts. For a prediction of the ISHLT grade (0R, 1R, 2/3R), AUROCs were 0.835, 0.633, and 0.905 in the cross-validated experiment and 0.764, 0.597, and 0.913; 0.631, 0.633, and 0.682; and 0.722, 0.601, and 0.805 in the validation cohorts, respectively. The predictions of the artificial intelligence model were interpretable by human experts and highlighted plausible morphological patterns.
Conclusion: We conclude that artificial intelligence can detect patterns of cellular transplant rejection in routine pathology, even when trained on small cohorts.
{"title":"Prediction of heart transplant rejection from routine pathology slides with self-supervised deep learning.","authors":"Tobias Paul Seraphin, Mark Luedde, Christoph Roderburg, Marko van Treeck, Pascal Scheider, Roman D Buelow, Peter Boor, Sven H Loosen, Zdenek Provaznik, Daniel Mendelsohn, Filip Berisha, Christina Magnussen, Dirk Westermann, Tom Luedde, Christoph Brochhausen, Samuel Sossalla, Jakob Nikolas Kather","doi":"10.1093/ehjdh/ztad016","DOIUrl":"https://doi.org/10.1093/ehjdh/ztad016","url":null,"abstract":"<p><strong>Aims: </strong>One of the most important complications of heart transplantation is organ rejection, which is diagnosed on endomyocardial biopsies by pathologists. Computer-based systems could assist in the diagnostic process and potentially improve reproducibility. Here, we evaluated the feasibility of using deep learning in predicting the degree of cellular rejection from pathology slides as defined by the International Society for Heart and Lung Transplantation (ISHLT) grading system.</p><p><strong>Methods and results: </strong>We collected 1079 histopathology slides from 325 patients from three transplant centres in Germany. We trained an attention-based deep neural network to predict rejection in the primary cohort and evaluated its performance using cross-validation and by deploying it to three cohorts. For binary prediction (rejection yes/no), the mean area under the receiver operating curve (AUROC) was 0.849 in the cross-validated experiment and 0.734, 0.729, and 0.716 in external validation cohorts. For a prediction of the ISHLT grade (0R, 1R, 2/3R), AUROCs were 0.835, 0.633, and 0.905 in the cross-validated experiment and 0.764, 0.597, and 0.913; 0.631, 0.633, and 0.682; and 0.722, 0.601, and 0.805 in the validation cohorts, respectively. The predictions of the artificial intelligence model were interpretable by human experts and highlighted plausible morphological patterns.</p><p><strong>Conclusion: </strong>We conclude that artificial intelligence can detect patterns of cellular transplant rejection in routine pathology, even when trained on small cohorts.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ftp.ncbi.nlm.nih.gov/pub/pmc/oa_pdf/33/45/ztad016.PMC10232288.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"9568841","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}
Ziliang Ye, Yanjun Zhang, Yuanyuan Zhang, Sisi Yang, Mengyi Liu, Qimeng Wu, Chun Zhou, Panpan He, Xiaoqin Gan, Xianhui Qin
Aims: The relationship between mobile phone use for making or receiving calls and hypertension risk remains uncertain. We aimed to examine the associations of mobile phone use for making or receiving calls and the use frequency with new-onset hypertension in the general population, using data from the UK Biobank.
Methods and results: A total of 212 046 participants without prior hypertension in the UK Biobank were included. Participants who have been using a mobile phone at least once per week to make or receive calls were defined as mobile phone users. The primary outcome was new-onset hypertension. During a median follow-up of 12.0 years, 13 984 participants developed new-onset hypertension. Compared with mobile phone non-users, a significantly higher risk of new-onset hypertension was found in mobile phone users [hazards ratio (HR), 1.07; 95% confidence interval (CI): 1.01-1.12]. Among mobile phone users, compared with those with a weekly usage time of mobile phones for making or receiving calls <5 mins, significantly higher risks of new-onset hypertension were found in participants with a weekly usage time of 30-59 mins (HR, 1.08; 95%CI: 1.01-1.16), 1-3 h (HR, 1.13; 95%CI: 1.06-1.22), 4-6 h (HR, 1.16; 95%CI: 1.04-1.29), and >6 h (HR, 1.25; 95%CI: 1.13-1.39) (P for trend <0.001). Moreover, participants with both high genetic risks of hypertension and longer weekly usage time of mobile phones making or receiving calls had the highest risk of new-onset hypertension.
Conclusions: Mobile phone use for making or receiving calls was significantly associated with a higher risk of new-onset hypertension, especially among high-frequency users.
{"title":"Mobile phone calls, genetic susceptibility, and new-onset hypertension: results from 212 046 UK Biobank participants.","authors":"Ziliang Ye, Yanjun Zhang, Yuanyuan Zhang, Sisi Yang, Mengyi Liu, Qimeng Wu, Chun Zhou, Panpan He, Xiaoqin Gan, Xianhui Qin","doi":"10.1093/ehjdh/ztad024","DOIUrl":"https://doi.org/10.1093/ehjdh/ztad024","url":null,"abstract":"<p><strong>Aims: </strong>The relationship between mobile phone use for making or receiving calls and hypertension risk remains uncertain. We aimed to examine the associations of mobile phone use for making or receiving calls and the use frequency with new-onset hypertension in the general population, using data from the UK Biobank.</p><p><strong>Methods and results: </strong>A total of 212 046 participants without prior hypertension in the UK Biobank were included. Participants who have been using a mobile phone at least once per week to make or receive calls were defined as mobile phone users. The primary outcome was new-onset hypertension. During a median follow-up of 12.0 years, 13 984 participants developed new-onset hypertension. Compared with mobile phone non-users, a significantly higher risk of new-onset hypertension was found in mobile phone users [hazards ratio (HR), 1.07; 95% confidence interval (CI): 1.01-1.12]. Among mobile phone users, compared with those with a weekly usage time of mobile phones for making or receiving calls <5 mins, significantly higher risks of new-onset hypertension were found in participants with a weekly usage time of 30-59 mins (HR, 1.08; 95%CI: 1.01-1.16), 1-3 h (HR, 1.13; 95%CI: 1.06-1.22), 4-6 h (HR, 1.16; 95%CI: 1.04-1.29), and >6 h (HR, 1.25; 95%CI: 1.13-1.39) (<i>P</i> for trend <0.001). Moreover, participants with both high genetic risks of hypertension and longer weekly usage time of mobile phones making or receiving calls had the highest risk of new-onset hypertension.</p><p><strong>Conclusions: </strong>Mobile phone use for making or receiving calls was significantly associated with a higher risk of new-onset hypertension, especially among high-frequency users.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ftp.ncbi.nlm.nih.gov/pub/pmc/oa_pdf/0c/7c/ztad024.PMC10232238.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"9566517","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}
Kai Ninomiya, Shigetaka Kageyama, Scot Garg, Shinichiro Masuda, Nozomi Kotoku, Pruthvi C Revaiah, Neil O'leary, Yoshinobu Onuma, Patrick W Serruys
Aims: Risk stratification and individual risk prediction play a key role in making treatment decisions in patients with complex coronary artery disease (CAD). The aim of this study was to assess whether machine learning (ML) algorithms can improve discriminative ability and identify unsuspected, but potentially important, factors in the prediction of long-term mortality following percutaneous coronary intervention or coronary artery bypass grafting in patients with complex CAD.
Methods and results: To predict long-term mortality, the ML algorisms were applied to the SYNTAXES database with 75 pre-procedural variables including demographic and clinical factors, blood sampling, imaging, and patient-reported outcomes. The discriminative ability and feature importance of the ML model was assessed in the derivation cohort of the SYNTAXES trial using a 10-fold cross-validation approach. The ML model showed an acceptable discrimination (area under the curve = 0.76) in cross-validation. C-reactive protein, patient-reported pre-procedural mental status, gamma-glutamyl transferase, and HbA1c were identified as important variables predicting 10-year mortality.
Conclusion: The ML algorithms disclosed unsuspected, but potentially important prognostic factors of very long-term mortality among patients with CAD. A 'mega-analysis' based on large randomized or non-randomized data, the so-called 'big data', may be warranted to confirm these findings.
{"title":"Can machine learning unravel unsuspected, clinically important factors predictive of long-term mortality in complex coronary artery disease? A call for 'big data'.","authors":"Kai Ninomiya, Shigetaka Kageyama, Scot Garg, Shinichiro Masuda, Nozomi Kotoku, Pruthvi C Revaiah, Neil O'leary, Yoshinobu Onuma, Patrick W Serruys","doi":"10.1093/ehjdh/ztad014","DOIUrl":"https://doi.org/10.1093/ehjdh/ztad014","url":null,"abstract":"<p><strong>Aims: </strong>Risk stratification and individual risk prediction play a key role in making treatment decisions in patients with complex coronary artery disease (CAD). The aim of this study was to assess whether machine learning (ML) algorithms can improve discriminative ability and identify unsuspected, but potentially important, factors in the prediction of long-term mortality following percutaneous coronary intervention or coronary artery bypass grafting in patients with complex CAD.</p><p><strong>Methods and results: </strong>To predict long-term mortality, the ML algorisms were applied to the SYNTAXES database with 75 pre-procedural variables including demographic and clinical factors, blood sampling, imaging, and patient-reported outcomes. The discriminative ability and feature importance of the ML model was assessed in the derivation cohort of the SYNTAXES trial using a 10-fold cross-validation approach. The ML model showed an acceptable discrimination (area under the curve = 0.76) in cross-validation. C-reactive protein, patient-reported pre-procedural mental status, gamma-glutamyl transferase, and HbA1c were identified as important variables predicting 10-year mortality.</p><p><strong>Conclusion: </strong>The ML algorithms disclosed unsuspected, but potentially important prognostic factors of very long-term mortality among patients with CAD. A 'mega-analysis' based on large randomized or non-randomized data, the so-called 'big data', may be warranted to confirm these findings.</p><p><strong>Clinical trial registration: </strong>SYNTAXES ClinicalTrials.gov reference: NCT03417050, SYNTAX ClinicalTrials.gov reference: NCT00114972.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ftp.ncbi.nlm.nih.gov/pub/pmc/oa_pdf/ac/da/ztad014.PMC10232230.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"9566521","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}
CardioPulse Digital talks to the Founder and Chief Medical Officer of AliveCor: Dr. David Albert David E.
{"title":"Meet key digital health thought leaders: David Albert.","authors":"Nico Bruining","doi":"10.1093/ehjdh/ztad020","DOIUrl":"https://doi.org/10.1093/ehjdh/ztad020","url":null,"abstract":"CardioPulse Digital talks to the Founder and Chief Medical Officer of AliveCor: Dr. David Albert David E.","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ftp.ncbi.nlm.nih.gov/pub/pmc/oa_pdf/a8/c8/ztad020.PMC10232255.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"9566523","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}
We recently published a novel categorization of studies related to virtual reality (VR) in your journal, European Heart Journal—Digital Health . 1 Our categorization is based on the usage of VR devices, where type A studies refer to those in which healthcare providers use VR devices and type B studies refer to those in which patients use them. Using this sim-ple definition, we clarified the study trends and characteristics of the two research directions. In this study, we used a classical natural language processing (NLP) methodology, specifically ‘term frequency– inverse document frequency’ to develop an automatic abstract categorizer, which is available as a web application at https://ahigaki-vr-categorizer-str-app-gb1m6v.streamlit.app
{"title":"ChatGPT's ability to classify virtual reality studies in cardiology.","authors":"Yuichiro Nakaya, Akinori Higaki, Osamu Yamaguchi","doi":"10.1093/ehjdh/ztad026","DOIUrl":"https://doi.org/10.1093/ehjdh/ztad026","url":null,"abstract":"We recently published a novel categorization of studies related to virtual reality (VR) in your journal, European Heart Journal—Digital Health . 1 Our categorization is based on the usage of VR devices, where type A studies refer to those in which healthcare providers use VR devices and type B studies refer to those in which patients use them. Using this sim-ple definition, we clarified the study trends and characteristics of the two research directions. In this study, we used a classical natural language processing (NLP) methodology, specifically ‘term frequency– inverse document frequency’ to develop an automatic abstract categorizer, which is available as a web application at https://ahigaki-vr-categorizer-str-app-gb1m6v.streamlit.app","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ftp.ncbi.nlm.nih.gov/pub/pmc/oa_pdf/3f/a5/ztad026.PMC10232268.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"9621354","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}
Vidhu Anand, Hanwen Hu, Alexander D Weston, Christopher G Scott, Hector I Michelena, Sorin V Pislaru, Rickey E Carter, Patricia A Pellikka
Aims: The current guidelines recommend aortic valve intervention in patients with severe aortic regurgitation (AR) with the onset of symptoms, left ventricular enlargement, or systolic dysfunction. Recent studies have suggested that we might be missing the window of early intervention in a significant number of patients by following the guidelines.
Methods and results: The overarching goal was to determine if machine learning (ML)-based algorithms could be trained to identify patients at risk for death from AR independent of aortic valve replacement (AVR). Models were trained with five-fold cross-validation on a dataset of 1035 patients, and performance was reported on an independent dataset of 207 patients. Optimal predictive performance was observed with a conditional random survival forest model. A subset of 19/41 variables was selected for inclusion in the final model. Variable selection was performed with 10-fold cross-validation using random survival forest model. The top variables included were age, body surface area, body mass index, diastolic blood pressure, New York Heart Association class, AVR, comorbidities, ejection fraction, end-diastolic volume, and end-systolic dimension, and the relative variable importance averaged across five splits of cross-validation in each repeat were evaluated. The concordance index for predicting survival of the best-performing model was 0.84 at 1 year, 0.86 at 2 years, and 0.87 overall, respectively.
Conclusion: Using common echocardiographic parameters and patient characteristics, we successfully trained multiple ML models to predict survival in patients with severe AR. This technique could be applied to identify high-risk patients who would benefit from early intervention, thereby improving patient outcomes.
{"title":"Machine learning-based risk stratification for mortality in patients with severe aortic regurgitation.","authors":"Vidhu Anand, Hanwen Hu, Alexander D Weston, Christopher G Scott, Hector I Michelena, Sorin V Pislaru, Rickey E Carter, Patricia A Pellikka","doi":"10.1093/ehjdh/ztad006","DOIUrl":"https://doi.org/10.1093/ehjdh/ztad006","url":null,"abstract":"<p><strong>Aims: </strong>The current guidelines recommend aortic valve intervention in patients with severe aortic regurgitation (AR) with the onset of symptoms, left ventricular enlargement, or systolic dysfunction. Recent studies have suggested that we might be missing the window of early intervention in a significant number of patients by following the guidelines.</p><p><strong>Methods and results: </strong>The overarching goal was to determine if machine learning (ML)-based algorithms could be trained to identify patients at risk for death from AR independent of aortic valve replacement (AVR). Models were trained with five-fold cross-validation on a dataset of 1035 patients, and performance was reported on an independent dataset of 207 patients. Optimal predictive performance was observed with a conditional random survival forest model. A subset of 19/41 variables was selected for inclusion in the final model. Variable selection was performed with 10-fold cross-validation using random survival forest model. The top variables included were age, body surface area, body mass index, diastolic blood pressure, New York Heart Association class, AVR, comorbidities, ejection fraction, end-diastolic volume, and end-systolic dimension, and the relative variable importance averaged across five splits of cross-validation in each repeat were evaluated. The concordance index for predicting survival of the best-performing model was 0.84 at 1 year, 0.86 at 2 years, and 0.87 overall, respectively.</p><p><strong>Conclusion: </strong>Using common echocardiographic parameters and patient characteristics, we successfully trained multiple ML models to predict survival in patients with severe AR. This technique could be applied to identify high-risk patients who would benefit from early intervention, thereby improving patient outcomes.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ftp.ncbi.nlm.nih.gov/pub/pmc/oa_pdf/41/f5/ztad006.PMC10232267.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"9571913","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 : 2023-04-11DOI: 10.1101/2023.04.09.23287650
A. Javed, D. Kim, S. Hershman, A. Shcherbina, Anderson Johnson, Alex Tolas, J. O’Sullivan, Michael V. McConnell, L. Lazzeroni, A. King, J. Christle, M. Oppezzo, C. Mattsson, Robert A. Harrington, M. Wheeler, Euan A Ashley
Background: Physical activity is strongly protective against the development of chronic diseases associated with aging. We previously demonstrated that digital interventions delivered through a smartphone app can increase short-term physical activity. Our randomized crossover trial has continued to digitally enroll participants, allowing increasing statistical power for greater precision in subsequent analyses. Methods: We offered enrollment to adults aged >=18 years with access to an iPhone and the MyHeart Counts app. After completion of a 1-week baseline period, e-consented participants were randomly allocated to four 7-day interventions. Interventions consisted of: 1) daily personalized e-coaching based on the individuals baseline activity patterns, 2) daily prompts to complete 10,000 steps, 3) hourly prompts to stand following inactivity, and 4) daily instructions to read guidelines from the American Heart Association website. The trial was completed in a free-living setting, where neither the participants or investigators were blinded to the intervention. The primary outcome was change in mean daily step count from baseline for each of the four interventions, assessed in a modified intention-to-treat analysis. This trial is registered with ClinicalTrials.gov, NCT03090321. Findings: Between January 1, 2017 and April 1, 2022, 4500 participants consented to enroll in the trial, of whom 2458 completed 7-days of baseline monitoring (mean daily steps 4232+/-73) and at least one day of one of the four interventions. The greater statistical power afforded by continued passive enrollment revealed that e-coaching prompts, tailored to an individual, increased step count significantly more than other interventions (402+/-71 steps, P=7.1x10-8). Interpretation: Digital studies can continuously recruit participants in a cost-effective manner, allowing for new insights provided by increased statistical power and refinement of prior signals. Here, we show that digital interventions tailored to an individual are effective in increasing short-term physical activity in a free-living cohort. Funding: Stanford Data Science Initiative and Catalyst Program, Apple, Google
{"title":"Personalized digital behaviour interventions increase short-term physical activity: a randomized control crossover trial substudy of the MyHeart Counts Cardiovascular Health Study","authors":"A. Javed, D. Kim, S. Hershman, A. Shcherbina, Anderson Johnson, Alex Tolas, J. O’Sullivan, Michael V. McConnell, L. Lazzeroni, A. King, J. Christle, M. Oppezzo, C. Mattsson, Robert A. Harrington, M. Wheeler, Euan A Ashley","doi":"10.1101/2023.04.09.23287650","DOIUrl":"https://doi.org/10.1101/2023.04.09.23287650","url":null,"abstract":"Background: Physical activity is strongly protective against the development of chronic diseases associated with aging. We previously demonstrated that digital interventions delivered through a smartphone app can increase short-term physical activity. Our randomized crossover trial has continued to digitally enroll participants, allowing increasing statistical power for greater precision in subsequent analyses. Methods: We offered enrollment to adults aged >=18 years with access to an iPhone and the MyHeart Counts app. After completion of a 1-week baseline period, e-consented participants were randomly allocated to four 7-day interventions. Interventions consisted of: 1) daily personalized e-coaching based on the individuals baseline activity patterns, 2) daily prompts to complete 10,000 steps, 3) hourly prompts to stand following inactivity, and 4) daily instructions to read guidelines from the American Heart Association website. The trial was completed in a free-living setting, where neither the participants or investigators were blinded to the intervention. The primary outcome was change in mean daily step count from baseline for each of the four interventions, assessed in a modified intention-to-treat analysis. This trial is registered with ClinicalTrials.gov, NCT03090321. Findings: Between January 1, 2017 and April 1, 2022, 4500 participants consented to enroll in the trial, of whom 2458 completed 7-days of baseline monitoring (mean daily steps 4232+/-73) and at least one day of one of the four interventions. The greater statistical power afforded by continued passive enrollment revealed that e-coaching prompts, tailored to an individual, increased step count significantly more than other interventions (402+/-71 steps, P=7.1x10-8). Interpretation: Digital studies can continuously recruit participants in a cost-effective manner, allowing for new insights provided by increased statistical power and refinement of prior signals. Here, we show that digital interventions tailored to an individual are effective in increasing short-term physical activity in a free-living cohort. Funding: Stanford Data Science Initiative and Catalyst Program, Apple, Google","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62358665","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-04-06eCollection Date: 2023-05-01DOI: 10.1093/ehjdh/ztad025
Le Li, Zhuxin Zhang, Likun Zhou, Zhenhao Zhang, Yulong Xiong, Zhao Hu, Yan Yao
Aims: Life-threatening ventricular arrhythmias (LTVAs) are common manifestations of sepsis. The majority of sepsis patients with LTVA are unresponsive to initial standard treatment and thus have a poor prognosis. There are very limited studies focusing on the early identification of patients at high risk of LTVA in sepsis to perform optimal preventive treatment interventions. We aimed to develop a prediction model to predict LTVA in sepsis using machine learning (ML) approaches.
Methods and results: Six ML algorithms including CatBoost, LightGBM, and XGBoost were employed to perform the model fitting. The least absolute shrinkage and selection operator (LASSO) regression was used to identify key features. Methods of model evaluation involved in this study included area under the receiver operating characteristic curve (AUROC), for model discrimination, calibration curve, and Brier score, for model calibration. Finally, we validated the prediction model both internally and externally. A total of 27 139 patients with sepsis were identified in this study, 1136 (4.2%) suffered from LTVA during hospitalization. We screened out 10 key features from the initial 54 variables via LASSO regression to improve the practicability of the model. CatBoost showed the best prediction performance among the six ML algorithms, with excellent discrimination (AUROC = 0.874) and calibration (Brier score = 0.157). The remarkable performance of the model was presented in the external validation cohort (n = 9492), with an AUROC of 0.836, suggesting certain generalizability of the model. Finally, a nomogram with risk classification of LTVA was shown in this study.
Conclusion: We established and validated a machine leaning-based prediction model, which was conducive to early identification of high-risk LTVA patients in sepsis, thus appropriate methods could be conducted to improve outcomes.
{"title":"Use of machine learning algorithms to predict life-threatening ventricular arrhythmia in sepsis.","authors":"Le Li, Zhuxin Zhang, Likun Zhou, Zhenhao Zhang, Yulong Xiong, Zhao Hu, Yan Yao","doi":"10.1093/ehjdh/ztad025","DOIUrl":"10.1093/ehjdh/ztad025","url":null,"abstract":"<p><strong>Aims: </strong>Life-threatening ventricular arrhythmias (LTVAs) are common manifestations of sepsis. The majority of sepsis patients with LTVA are unresponsive to initial standard treatment and thus have a poor prognosis. There are very limited studies focusing on the early identification of patients at high risk of LTVA in sepsis to perform optimal preventive treatment interventions. We aimed to develop a prediction model to predict LTVA in sepsis using machine learning (ML) approaches.</p><p><strong>Methods and results: </strong>Six ML algorithms including CatBoost, LightGBM, and XGBoost were employed to perform the model fitting. The least absolute shrinkage and selection operator (LASSO) regression was used to identify key features. Methods of model evaluation involved in this study included area under the receiver operating characteristic curve (AUROC), for model discrimination, calibration curve, and Brier score, for model calibration. Finally, we validated the prediction model both internally and externally. A total of 27 139 patients with sepsis were identified in this study, 1136 (4.2%) suffered from LTVA during hospitalization. We screened out 10 key features from the initial 54 variables via LASSO regression to improve the practicability of the model. CatBoost showed the best prediction performance among the six ML algorithms, with excellent discrimination (AUROC = 0.874) and calibration (Brier score = 0.157). The remarkable performance of the model was presented in the external validation cohort (<i>n</i> = 9492), with an AUROC of 0.836, suggesting certain generalizability of the model. Finally, a nomogram with risk classification of LTVA was shown in this study.</p><p><strong>Conclusion: </strong>We established and validated a machine leaning-based prediction model, which was conducive to early identification of high-risk LTVA patients in sepsis, thus appropriate methods could be conducted to improve outcomes.</p>","PeriodicalId":72965,"journal":{"name":"European heart journal. Digital health","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ftp.ncbi.nlm.nih.gov/pub/pmc/oa_pdf/b8/c8/ztad025.PMC10232270.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"9568846","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}