Summary of Predictive Modeling with Temporal Graphical Representation on Electronic Health Records, by Jiayuan Chen et al.
Predictive Modeling with Temporal Graphical Representation on Electronic Health Records
by Jiayuan Chen, Changchang Yin, Yuanlong Wang, Ping Zhang
First submitted to arxiv on: 7 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 This paper proposes a novel deep learning-based predictive model for healthcare using Electronic Health Records (EHR). It aims to develop an effective representation of a patient’s EHR by hierarchically incorporating both temporal relationships between historical visits and medical events, as well as structural information within these elements. The authors critique existing methods, which either focus on sequential representations or graphical representations, and argue that their proposed approach can capture both types of information. They model a patient’s EHR as a novel temporal heterogeneous graph, integrating structured information from medical event nodes to visit nodes and utilizing time-aware visit nodes to capture changes in the patient’s health status. The authors also introduce a novel temporal graph transformer (TRANS) that combines temporal edge features, global positional encoding, and local structural encoding into heterogeneous graph convolution. They validate the effectiveness of TRANS through extensive experiments on three real-world datasets, achieving state-of-the-art performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper uses artificial intelligence to help doctors predict what might happen to a patient based on their medical history. It’s like trying to figure out what a person might do next based on all the things they’ve done in the past. The researchers created a special kind of graph, like a map, that shows how different events and visits are connected over time. They also developed a new way to look at this graph, called TRANS, which helps doctors understand how these connections change over time. By using real patient data, the team showed that their approach works better than other methods. |
Keywords
» Artificial intelligence » Deep learning » Positional encoding » Transformer
