Search

Your search keyword '"Pomerantsev, Eugene"' showing total 276 results
Start Over Author "Pomerantsev, Eugene"
276 results on '"Pomerantsev, Eugene"'

1. Signature Activation: A Sparse Signal View for Holistic Saliency

Author

Ayala, Jose Roberto Tello, Fahed, Akl C., Pan, Weiwei, Pomerantsev, Eugene V., Ellinor, Patrick T., Philippakis, Anthony, and Doshi-Velez, Finale

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Machine Learning
Abstract

The adoption of machine learning in healthcare calls for model transparency and explainability. In this work, we introduce Signature Activation, a saliency method that generates holistic and class-agnostic explanations for Convolutional Neural Network (CNN) outputs. Our method exploits the fact that certain kinds of medical images, such as angiograms, have clear foreground and background objects. We give theoretical explanation to justify our methods. We show the potential use of our method in clinical settings through evaluating its efficacy for aiding the detection of lesions in coronary angiograms.

Published
2023

2. Data Augmentation for Electrocardiograms

Author

Raghu, Aniruddh, Shanmugam, Divya, Pomerantsev, Eugene, Guttag, John, and Stultz, Collin M.

Subjects
Computer Science - Machine Learning
Abstract

Neural network models have demonstrated impressive performance in predicting pathologies and outcomes from the 12-lead electrocardiogram (ECG). However, these models often need to be trained with large, labelled datasets, which are not available for many predictive tasks of interest. In this work, we perform an empirical study examining whether training time data augmentation methods can be used to improve performance on such data-scarce ECG prediction problems. We investigate how data augmentation strategies impact model performance when detecting cardiac abnormalities from the ECG. Motivated by our finding that the effectiveness of existing augmentation strategies is highly task-dependent, we introduce a new method, TaskAug, which defines a flexible augmentation policy that is optimized on a per-task basis. We outline an efficient learning algorithm to do so that leverages recent work in nested optimization and implicit differentiation. In experiments, considering three datasets and eight predictive tasks, we find that TaskAug is competitive with or improves on prior work, and the learned policies shed light on what transformations are most effective for different tasks. We distill key insights from our experimental evaluation, generating a set of best practices for applying data augmentation to ECG prediction problems., Comment: Conference on Health, Inference, and Learning (CHIL) 2022

Published
2022

4. Learning to Predict with Supporting Evidence: Applications to Clinical Risk Prediction

Author

Raghu, Aniruddh, Guttag, John, Young, Katherine, Pomerantsev, Eugene, Dalca, Adrian V., and Stultz, Collin M.

Subjects
Computer Science - Machine Learning
Abstract

The impact of machine learning models on healthcare will depend on the degree of trust that healthcare professionals place in the predictions made by these models. In this paper, we present a method to provide people with clinical expertise with domain-relevant evidence about why a prediction should be trusted. We first design a probabilistic model that relates meaningful latent concepts to prediction targets and observed data. Inference of latent variables in this model corresponds to both making a prediction and providing supporting evidence for that prediction. We present a two-step process to efficiently approximate inference: (i) estimating model parameters using variational learning, and (ii) approximating maximum a posteriori estimation of latent variables in the model using a neural network, trained with an objective derived from the probabilistic model. We demonstrate the method on the task of predicting mortality risk for patients with cardiovascular disease. Specifically, using electrocardiogram and tabular data as input, we show that our approach provides appropriate domain-relevant supporting evidence for accurate predictions., Comment: ACM Conference on Health, Learning, and Inference 2021

Published
2021

12. Metabolite Profiles Predict Acute Kidney Injury and Mortality in Patients Undergoing Transcatheter Aortic Valve Replacement

Author

Elmariah, Sammy, Farrell, Laurie A, Daher, Maureen, Shi, Xu, Keyes, Michelle J, Cain, Carolyn H, Pomerantsev, Eugene, Vlahakes, Gus J, Inglessis, Ignacio, Passeri, Jonathan J, Palacios, Igor F, Fox, Caroline S, Rhee, Eugene P, and Gerszten, Robert E

Subjects
Transplantation, Kidney Disease, Clinical Research, Aging, Heart Disease, Cardiovascular, 2.1 Biological and endogenous factors, Aetiology, Renal and urogenital, Good Health and Well Being, Acute Kidney Injury, Aged, Aged, 80 and over, Aortic Valve Stenosis, Biomarkers, Cardiac Catheterization, Chromatography, Liquid, Female, Heart Valve Prosthesis Implantation, Humans, Kaplan-Meier Estimate, Linear Models, Logistic Models, Male, Mass Spectrometry, Massachusetts, Metabolomics, Multivariate Analysis, Odds Ratio, Predictive Value of Tests, Prevalence, Risk Assessment, Risk Factors, S-Adenosylhomocysteine, Treatment Outcome, aortic stenosis, kidney, metabolomics, mortality, transcatheter aortic valve implantation, Cardiorespiratory Medicine and Haematology
Abstract

BackgroundAcute kidney injury (AKI) occurs commonly after transcatheter aortic valve replacement (TAVR) and is associated with markedly increased postoperative mortality. We previously identified plasma metabolites predictive of incident chronic kidney disease, but whether metabolite profiles can identify those at risk of AKI is unknown.Methods and resultsWe performed liquid chromatography-mass spectrometry-based metabolite profiling on plasma from patients undergoing TAVR and subjects from the community-based Framingham Heart Study (N=2164). AKI was defined by using the Valve Academic Research Consortium-2 criteria. Of 44 patients (mean age 82±9 years, 52% female) undergoing TAVR, 22 (50%) had chronic kidney disease and 9 (20%) developed AKI. Of 85 metabolites profiled, we detected markedly concordant cross-sectional metabolic changes associated with chronic kidney disease in the hospital-based TAVR and Framingham Heart Study cohorts. Baseline levels of 5-adenosylhomocysteine predicted AKI after TAVR, despite adjustment for baseline glomerular filtration rate (odds ratio per 1-SD increase 5.97, 95% CI 1.62-22.0; P=0.007). Of the patients who had AKI, 6 (66.7%) subsequently died, compared with 3 (8.6%) deaths among those patients who did not develop AKI (P=0.0008) over a median follow-up of 7.8 months. 5-adenosylhomocysteine was predictive of all-cause mortality after TAVR (hazard ratio per 1-SD increase 2.96, 95% CI 1.33-6.58; P=0.008), independent of baseline glomerular filtration rate.ConclusionsIn an elderly population with severe aortic stenosis undergoing TAVR, metabolite profiling improves the prediction of AKI. Given the multifactorial nature of AKI after TAVR, metabolite profiles may identify those patients with reduced renal reserve.

Published
2016

14. Identifying high risk clinical phenogroups of pulmonary hypertension through a clustering analysis

Author

Rambarat, Paula, primary, Zern, Emily K., additional, Wang, Dongyu, additional, Roshandelpoor, Athar, additional, Zarbafian, Shahrooz, additional, Liu, Elizabeth E., additional, Wang, Jessica K., additional, McNeill, Jenna N., additional, Andrews, Carl T., additional, Pomerantsev, Eugene V., additional, Diamant, Nathaniel, additional, Batra, Puneet, additional, Lubitz, Steven A., additional, Picard, Michael H., additional, and Ho, Jennifer E., additional

Published
2023
Full Text
View/download PDF

20. Long-Term Experience and Outcomes With Transcatheter Closure of Patent Foramen Ovale

Author

Inglessis, Ignacio, Elmariah, Sammy, Rengifo-Moreno, Pablo A., Margey, Ronan, O'Callaghan, Caitlin, Cruz-Gonzalez, Ignacio, Baron, Suzanne, Mehrotra, Praveen, Tan, Timothy C., Hung, Judy, Demirjian, Zareh N., Buonanno, Ferdinando S., Ning, MingMing, Silverman, Scott B., Cubeddu, Roberto J., Pomerantsev, Eugene, Schainfeld, Robert M., Dec, G. William, Jr., and Palacios, Igor F.

Published
2013
Full Text
View/download PDF

25. Abstract 10389: Identifying High-Risk Clinical Phenogroups of Pulmonary Hypertension Through a Clustering Analysis

Author

Rambarat, Paula, primary, Zern, Emily K, additional, Wang, Dongyu, additional, Zarbafian, Shahrooz, additional, Liu, Elizabeth E, additional, Wang, Jessica K, additional, McNeill, Jenna N, additional, Andrews, Carl T, additional, Pomerantsev, Eugene V, additional, Diamant, Nathaniel, additional, Batra, Puneet, additional, Lubitz, Steven A, additional, Picard, Michael H, additional, and Ho, Jennifer E, additional

Published
2021
Full Text
View/download PDF

26. Abstract 12320: Long-Term Outcomes of Open and Endovascular Management of Adult Aortic Coarctation

Author

Moonsamy, Philicia, primary, Bhatt, Ami B, additional, Sundt, Thoralf M, additional, Defaria Yeh, Doreen, additional, Sakhuja, Rahul, additional, Boltje, Jan W, additional, Pagliaro, Jaclyn, additional, Pomerantsev, Eugene V, additional, Bloom, Jordan P, additional, Li, Selena, additional, Cameron, Duke E, additional, Eagleton, Matthew, additional, and Jassar, Arminder, additional

Published
2021
Full Text
View/download PDF

34. Comparison of Sixty-Four–Slice Multidetector Computed Tomographic Coronary Angiography to Coronary Angiography With Intravascular Ultrasound for the Detection of Transplant Vasculopathy

Author

Gregory, Shawn A., Ferencik, Maros, Achenbach, Stephan, Yeh, Robert W., Hoffmann, Udo, Inglessis, Ignacio, Cury, Ricardo C., Nieman, Koen, McNulty, Iris A., Laffan, Josephine A., Pomerantsev, Eugene V., Brady, Thomas J., Semigran, Marc J., and Jang, Ik-Kyung

Published
2006
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results