Columbia Technology Ventures

Artificial intelligence-based ECG analysis for comprehensive heart disease detection

This technology is a deep learning model, termed EchoNext, that uses standard electrocardiogram (ECG) data to predict multiple forms of structural heart disease.

Unmet Need: Accessible, low-cost methods for comprehensive heart disease detection

Current methods for diagnosing structural heart diseases rely heavily on imaging techniques like echocardiography, which can be time-consuming, costly, and not readily accessible in all clinical settings. Although low-cost and accessible, standard electrocardiogram (ECG) analysis is limited in its ability to detect a full spectrum of heart diseases, often missing conditions like ventricular dysfunction and valvular diseases. There is a need for a more comprehensive, accessible, and efficient method to identify patients with latent structural heart diseases who would benefit from further diagnostic evaluation.

The Technology: Deep learning ECG model for comprehensive heart disease screening

This technology utilizes advanced deep learning algorithms to analyze standard 12-lead ECG data, along with patient age, sex, and key ECG-derived measurements, to predict the presence of various structural heart diseases. By detecting subtle patterns in ECG signals indicative of conditions such as ventricular dysfunction, increased wall thickness, valvular diseases, pericardial effusion, and pulmonary hypertension, the model provides a comprehensive assessment from a single ECG test.

This technology was developed using a dataset of over 1 million ECGs.

Applications:

  • Clinical screening tool for early detection of structural heart diseases
  • Automated ECG analysis for referral decisions to echocardiography
  • Integration into ECG devices or software systems for hospitals and clinics
  • Enhancement of clinical workflows and decision-making in cardiology practices
  • Population health screening in remote or resource-limited settings
  • Research tool for studying ECG-based disease patterns

Advantages:

  • Comprehensive detection of multiple heart diseases from a single ECG test
  • Superior precision compared to models targeting individual conditions
  • Robust performance across diverse patient populations and clinical settings
  • Enhances clinical workflow by identifying patients needing further imaging
  • Facilitates earlier disease detection and management
  • Cost-effective and accessible compared to imaging-based diagnostics

Lead Inventor:

Pierre Elias, MD

Patent Information:

Patent Pending

Related Publications:

Tech Ventures Reference: