Multimodal ECG Fusion for Pulmonary Embolism Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current diagnostic methods for pulmonary embolism (PE) rely heavily on thoracic imaging modalities like CTPA, which are resource-intensive and expose patients to radiation, with low specificity and high diagnostic uncertainty.
Innovation Solution
A deep learning-based multimodal fusion model that integrates ECG waveform data and discrete patient data to predict the likelihood of PE, using supervised learning to optimize the match with CTPA results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If CTPA scans are used to confirm PE diagnosis, then diagnostic accuracy is improved, but resource utilization increases and patient radiation exposure increases
Solution Approach 1:
The system performs preliminary risk stratification using ECG and clinical data before ordering CTPA scans. By pre-identifying low-risk patients through the deep learning model, unnecessary CTPA scans are avoided, reducing radiation exposure while maintaining diagnostic accuracy for those who truly need it.
Solution Approach 2:
The deep learning model acts as an intermediary between initial clinical assessment and CTPA scanning. It processes ECG waveform data and clinical variables to generate a risk score, serving as a filtering layer that determines which patients proceed to CTPA, thereby reducing overall radiation exposure while preserving diagnostic accuracy.
2Measurement precision
If CTPA scans are used to confirm PE diagnosis, then diagnostic accuracy is improved, but resource utilization increases
Solution Approach 1:
The system performs preliminary risk stratification using ECG and clinical data before ordering CTPA scans. By pre-identifying low-risk patients through the deep learning model, unnecessary CTPA scans are avoided, reducing resource utilization while maintaining diagnostic accuracy for those who truly need it.
Solution Approach 2:
The deep learning model acts as an intermediary between initial clinical assessment and CTPA scanning. It processes ECG waveform data and clinical variables to generate a risk score, serving as a filtering layer that determines which patients proceed to CTPA, thereby reducing overall resource utilization while preserving diagnostic accuracy.
3Productivity
If clinical decision rules are used for PE assessment, then resource utilization is reduced, but diagnostic uncertainty increases
Solution Approach 1:
The system replaces traditional mechanical clinical decision rules with an intelligent deep learning model that processes ECG waveform data. This substitution reduces diagnostic uncertainty by capturing complex patterns in ECG data that traditional rules miss, while also reducing resource utilization by accurately identifying low-risk patients.
Solution Approach 2:
The system transforms the assessment approach from discrete clinical rule parameters to continuous ECG waveform analysis. By extracting temporal and spectral features from ECG signals, the model achieves higher diagnostic reliability while reducing unnecessary CTPA scans, thus lowering resource utilization.
Data Source
AI summary
A method of assessing a likelihood of a patient having a pulmonary embolism (PE) comprises receiving discrete patient data, including patient-related clinical and demographic data pertinent to the patient, receiving electrocardiograph (ECG) waveform data obtained from examination of the patient, processing both the received discrete patient data and received set of ECG waveform data using a supervised deep learning multimodal fusion model that has been trained using analogous input training data including both discrete patient data and ECG waveform data to obtain an optimal match to results from corresponding patient computed tomography pulmonary angiograms (CTPA), indicative of a presence or absence of a pulmonary embolism, and outputting a measure of the likelihood of the patient having a pulmonary embolism.