Early Prediction of Cardiomyopathy in Pediatric Cancer Survivors Using Echocardiographic Video-Based Deep Learning.
13 affiliations
- 1Department of Bioengineering, University of Washington, Seattle, Washington, USA.
- 2Seattle Children's Hospital, Seattle, Washington, USA.
- 3Departments of Pediatrics and Population Sciences, City of Hope, Duarte, California, USA.
- 4Department Of Pediatrics, University of Washington, Seattle, Washington, USA.
- 5Clinical Research and Public Health Sciences Divisions, Fred Hutchinson Cancer Center, Seattle, Washington, USA.
- 6Department Of Pediatrics, Emory University, Atlanta, Georgia, USA.
- 7Aflac Cancer and Blood Disorders Center of Children's Healthcare of Atlanta and Department of Pediatrics, Emory University, Atlanta, Georgia, USA.
- 8Department Of Pediatrics, University of Minnesota, Masonic Children's Hospital, Minneapolis, Minnesota, USA.
- 9Department Of Pediatrics, University of Toronto, The Hospital for Sick Children, Toronto, Ontario, Canada.
- 10Departments Of Pediatrics, Duke University Medical Center, Durham, North Carolina, USA.
- 11Division of Cardiology, University of Washington, Seattle, Washington, USA.
- 12Institute For Stem Cell and Regenerative Medicine, University of Washington, Seattle, Washington, USA.
- 13Center For Cardiovascular Biology, University of Washington, Seattle, Washington, USA.
Abstract
Purpose
Cardiomyopathy (CM) is a leading cause of late mortality in childhood cancer survivors, but it is often identified only after irreversible cardiac remodeling has already occurred. In this proof-of-concept study, we assessed whether deep learning applied to routine echocardiographic videos can identify features predictive of future CM.Methods
We analyzed parasternal short-axis clips from cancer survivors ≤21 years old at cancer diagnosis. 3D deep convolutional neural networks were trained to classify CM-positive versus CM-negative studies from echocardiograms conducted up to five years prior to CM diagnosis. The performance was assessed on internal hold-out validation (Center 1) and external validation cohorts (Centers 2, 3). The model's prognostic value was evaluated by stratifying patients into high- and low-risk groups for developing CM, and probability of future CM was assessed using cumulative incidence and proportional hazards regression analysis.Results
The model achieved an area under the receiver operating characteristic curve (AUROC) of 0.87±0.05 and an area under the precision-recall curve (AUPRC) of 0.74±0.07 on the internal hold-out set. External validation showed near-chance single-study discrimination (AUROC 0.55±0.02 and 0.60±0.03; AUPRC 0.22±0.02 and 0.21±0.03 for Centers 2 and 3, respectively). Nevertheless, analysis of the external cohort showed significantly higher probability of CM in high- versus low-risk groups (hazard ratio 2.78, 95% CI = 1.52-5.10, p = 0.00056).Conclusion
Deep learning models have the potential to identify subtle echocardiographic markers years before CM diagnosis in pediatric cancer survivors and could serve as a preliminary step toward risk-guided CM surveillance. Further refinement and validation are needed to determine the utility of deep learning for early identification of pathologic cardiac remodeling in this population.
Related
- Contrasting Echocardiographic Trajectories with and Without Tafamidis in Wild-Type Transthyretin Amyloid Cardiomyopathy: A Real-World Observational Study with Two-Dimensional Speckle-Tracking Analysis
- Atypical echocardiographic phenotype of end-stage hypertrophic cardiomyopathy: a case series
- HCM-PhenotypeNet: Deep Multimodal Phenotyping of Hypertrophic Cardiomyopathy from Echocardiographic Video and Clinical Data
- 179 Machine learning-based prediction of CAR-T cell therapy response in pediatric acute lymphoblastic leukemia patients using single-cell multi-omics features
- Machine learning prediction model for distant metastasis at diagnosis in pediatric neuroblastoma: development, interpretation, and temporal validation
BibTeX
@article{sharma2026early,
title = {{Early Prediction of Cardiomyopathy in Pediatric Cancer Survivors Using Echocardiographic Video-Based Deep Learning.}},
author = {Surbhi Sharma and Brandy Martinez Di Pietro and Ananya Nandula and Tanishka Goplani and Hanpeng Chen and Saro Armenian and Aarti Bhat and Nancy Blythe and William Border and Kasey J Leger and Wendy M Leisenring and Lillian R Meacham and Shanti Narasimhan and Paul C Nathan and Ritu Sachdeva and Karim Sadak and Kayla Stratton and Eric J Chow and Lindsay A Edwards and Patrick M Boyle},
journal = {Echocardiography (Mount Kisco, N.Y.)},
year = {2026},
month = oct,
doi = {10.1111/echo.70643},
url = {https://europepmc.org/article/MED/42908394},
note = {CC BY},
}
RIS
TY - JOUR
TI - Early Prediction of Cardiomyopathy in Pediatric Cancer Survivors Using Echocardiographic Video-Based Deep Learning.
AU - Surbhi Sharma
AU - Brandy Martinez Di Pietro
AU - Ananya Nandula
AU - Tanishka Goplani
AU - Hanpeng Chen
AU - Saro Armenian
AU - Aarti Bhat
AU - Nancy Blythe
AU - William Border
AU - Kasey J Leger
AU - Wendy M Leisenring
AU - Lillian R Meacham
AU - Shanti Narasimhan
AU - Paul C Nathan
AU - Ritu Sachdeva
AU - Karim Sadak
AU - Kayla Stratton
AU - Eric J Chow
AU - Lindsay A Edwards
AU - Patrick M Boyle
PY - 2026
DA - 2026/10/01
JO - Echocardiography (Mount Kisco, N.Y.)
DO - 10.1111/echo.70643
UR - https://europepmc.org/article/MED/42908394
AB - Purpose Cardiomyopathy (CM) is a leading cause of late mortality in childhood cancer survivors, but it is often identified only after irreversible cardiac remodeling has already occurred. In this proof-of-concept study, we assessed whether deep learning applied to routine echocardiographic videos can identify features predictive of future CM.Methods We analyzed parasternal short-axis clips from cancer survivors ≤21 years old at cancer diagnosis. 3D deep convolutional neural networks were trained to classify CM-positive versus CM-negative studies from echocardiograms conducted up to five years prior to CM diagnosis. The performance was assessed on internal hold-out validation (Center 1) and external validation cohorts (Centers 2, 3). The model's prognostic value was evaluated by stratifying patients into high- and low-risk groups for developing CM, and probability of future CM was assessed using cumulative incidence and proportional hazards regression analysis.Results The model achieved an area under the receiver operating characteristic curve (AUROC) of 0.87±0.05 and an area under the precision-recall curve (AUPRC) of 0.74±0.07 on the internal hold-out set. External validation showed near-chance single-study discrimination (AUROC 0.55±0.02 and 0.60±0.03; AUPRC 0.22±0.02 and 0.21±0.03 for Centers 2 and 3, respectively). Nevertheless, analysis of the external cohort showed significantly higher probability of CM in high- versus low-risk groups (hazard ratio 2.78, 95% CI = 1.52-5.10, p = 0.00056).Conclusion Deep learning models have the potential to identify subtle echocardiographic markers years before CM diagnosis in pediatric cancer survivors and could serve as a preliminary step toward risk-guided CM surveillance. Further refinement and validation are needed to determine the utility of deep learning for early identification of pathologic cardiac remodeling in this population.
SN - 0742-2822
SN - 1540-8175
ER -