Summary of Mimic-iv-ext-pe: Using a Large Language Model to Predict Pulmonary Embolism Phenotype in the Mimic-iv Dataset, by B. D. Lam et al.
MIMIC-IV-Ext-PE: Using a large language model to predict pulmonary embolism phenotype in the MIMIC-IV dataset
by B. D. Lam, S. Ma, I. Kovalenko, P. Wang, O. Jafari, A. Li, S. Horng
First submitted to arxiv on: 29 Oct 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Machine Learning (cs.LG)
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 The paper presents an AI-based approach to automatically label computed tomography pulmonary angiography (CTPA) scans for pulmonary embolism (PE), leveraging a pre-trained Bio_ClinicalBERT transformer language model, VTE-BERT. The authors used the MIMIC-IV database to extract radiology reports and manually labeled them as PE-positive or PE-negative by two physicians. They then applied VTE-BERT to automatically extract labels, verifying its reliability through manual adjudication. The results show that VTE-BERT achieves a sensitivity of 92.4% and positive predictive value (PPV) of 87.8% on all patients with CTPA radiology reports from the emergency room or hospital admission. This is compared to diagnosis codes, which have a sensitivity of 95.4% and PPV of 83.8% on hospitalized patients with discharge diagnosis codes. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper uses AI to help doctors identify pulmonary embolism (PE) in CT scans more accurately. They took data from a big hospital database and used it to train a special computer program called VTE-BERT. This program can look at CT scan reports and automatically say whether they show PE or not. The researchers tested VTE-BERT against what doctors had already said, and found that it was very good at getting the right answer. They also compared it to how doctors diagnosed PE in the past, and showed that VTE-BERT is almost as good. |
Keywords
» Artificial intelligence » Bert » Language model » Transformer
