1)Goto S et al. Artificial intelligence to predict needs for urgent revascularization from 12-leads electrocardiography in emergency patients. PLoS ONE. 2019;14:e0210103
A study that created an AI that highly accurately determines whether catheter treatment is necessary based on electrocardiograms obtained during emergency outpatient visits. This is the world's first research in which AI is trained to use 12-lead electrocardiograms as time-series potential data, and it formed the basis for subsequent AI research using electrocardiograms.
2)Goto S et al. Artificial intelligence-enabled fully automated detection of cardiac amyloidosis using electrocardiograms and echocardiograms. Nature Communication. 2021;12:2726.
Using electrocardiograms and echocardiograms collected from five facilities in Japan and the United States, AI was trained to create a model that detects cardiac amyloidosis. The model showed that by combining electrocardiogram and echocardiography, it was possible to detect cardiac amyloidosis with a positive predictive value of over 60%.
3)Goto S, Yagi R et al. Multinational Federated Learning Approach to Train ECG and Echocardiogram Models for Hypertrophic Cardiomyopathy Detection. Circulation. 2021;CIRCULATIONAHA121058696.
This paper examines the accuracy of detecting hypertrophic cardiomyopathy by using AI to learn in conjunction with electrocardiograms and echocardiograms collected from four facilities in Japan and the United States. We showed that AI trained on single-center data does not necessarily guarantee generalization performance for multi-center data, and that learning data from multiple facilities through federated learning improves accuracy and generalization performance. This is the world's first paper applying federated learning to electrocardiograms and echocardiograms.
4)Yagi R, Goto S et al. Importance of external validation and subgroup analysis of artificial intelligence in the detection of low ejection fraction from electrocardiograms. European Heart Journal - Digital Health. 2022;3:4
This research created an AI that detects patients with decreased heart contractility using electrocardiogram data collected from four facilities in Japan and the United States. AI that trained on data from a single facility using an approach that trained on data from its own facility and evaluated generalization performance at the remaining three facilities showed that although it performed well with data from its own facility, performance deteriorated when using data from other facilities.
5)A Gearhart, S Goto (equal contribution) RC Deo, AJ Powell. An Automated View Classification Model for Pediatric Echocardiography Using Artificial Intelligence. J Am Soc Echocardiogr. 2022 Aug 29:S0894-7317(22)00428-X. doi : 10.1016/j.echo.2022.08.009. Epub ahead of print. PMID: 36049595.
We created an AI that learns echocardiographic data and automatically determines echocardiographic cross-sections. This model was able to determine echocardiographic sections with an accuracy of over 90%. It is expected to be a basic technology necessary for automating various echocardiographic analyses.
6)Yagi R , Goto S et al. Expanded adaptation of an artificial intelligence model for predicting chemotherapy-induced cardiotoxicity using baseline electrocardiograms. European Society of Cardiology Congress, 2022, Barcelona, Spain
Research that built an AI that predicts.
7)Yagi R , Goto S et al. Stratification of the Risk of Developing Heart Failure in Patients With Left Bundle Branch Block: Approach Using Artificial Intelligence. American Heart Association Scientific Session, 2022, Chicago, USA
This study showed that there is heterogeneity in the population of people with heart failure, and that AI that has learned a large number of electrocardiograms can identify patients at high risk of developing heart failure.
8)Miura K, Yagi R (equal contribution), Goto S et al. Deep Learning-Based Model Detects Atrial Septal Defects from Electrocardiography: A Cross-sectional Multicenter Hospital-Based Study. eClinical Medicine. 2023. in press
A study that developed an AI that detects atrial septal defect from electrocardiograms collected from three facilities in Japan and the United States. AI that has learned a large number of electrocardiograms can perform her task with high accuracy, which has been considered difficult even for experienced clinicians, to ``detect atrial septal defect from electrocardiograms alone.'' The paper shown.