Summary of Multimodal Variational Autoencoder For Low-cost Cardiac Hemodynamics Instability Detection, by Mohammod N. I. Suvon et al.
Multimodal Variational Autoencoder for Low-cost Cardiac Hemodynamics Instability Detection
by Mohammod N. I. Suvon, Prasun C. Tripathi, Wenrui Fan, Shuo Zhou, Xianyuan Liu, Samer Alabed, Venet Osmani, Andrew J. Swift, Chen Chen, Haiping Lu
First submitted to arxiv on: 20 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computer Vision and Pattern Recognition (cs.CV)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary Recent advancements in cardiac hemodynamic instability (CHDI) detection have primarily focused on single-data-modality machine learning approaches using cardiac magnetic resonance imaging (MRI). However, these methods often fall short due to limited labeled patient data, a common challenge in the medical domain. Additionally, few studies have explored multimodal methods for CHDI detection, relying on costly modalities like MRI and echocardiogram. In response, we propose CardioVAE_XG, a novel multimodal variational autoencoder integrating low-cost chest X-ray (CXR) and electrocardiogram (ECG) modalities with pre-training on a large unlabeled dataset. CardioVAE_XG introduces a tri-stream pre-training strategy to learn shared and modality-specific features, enabling fine-tuning with both unimodal and multimodal datasets. We pre-trained the model on 50,982 subjects from MIMIC database and fine-tuned it on 795 subjects from ASPIRE registry. Evaluations show CardioVAE_XG offers promising performance (AUROC = 0.79 and Accuracy = 0.77) in non-invasive CHDI prediction, representing a significant step forward. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about developing a new way to predict when someone might have cardiac problems just by looking at their chest X-ray and electrocardiogram (ECG). Right now, doctors mainly use expensive tests like MRI to detect these problems. But this new method uses cheaper and more accessible tests, making it easier for doctors to diagnose cardiac issues early on. The new approach is called CardioVAE_XG, and it’s really good at predicting when someone might have a cardiac problem (it gets 0.79 out of 1 in accuracy). This can help doctors make better decisions about treatment and care. |
Keywords
* Artificial intelligence * Fine tuning * Machine learning * Variational autoencoder
