Automated Extraction of VTE Events From Narrative Radiology Reports in Electronic Health Records
Bibliographic record
Abstract
BACKGROUND: Surveillance of venous thromboembolisms (VTEs) is necessary for improving patient safety in acute care hospitals, but current detection methods are inaccurate and inefficient. With the growing availability of clinical narratives in an electronic format, automated surveillance using natural language processing (NLP) techniques may represent a better method. OBJECTIVE: We assessed the accuracy of using symbolic NLP for identifying the 2 clinical manifestations of VTE, deep vein thrombosis (DVT) and pulmonary embolism (PE), from narrative radiology reports. METHODS: A random sample of 4000 narrative reports was selected among imaging studies that could diagnose DVT or PE, and that were performed between 2008 and 2012 in a university health network of 5 adult-care hospitals in Montreal (Canada). The reports were coded by clinical experts to identify positive and negative cases of DVT and PE, which served as the reference standard. Using data from the largest hospital (n=2788), 2 symbolic NLP classifiers were trained; one for DVT, the other for PE. The accuracy of these classifiers was tested on data from the other 4 hospitals (n=1212). RESULTS: On manual review, 663 DVT-positive and 272 PE-positive reports were identified. In the testing dataset, the DVT classifier achieved 94% sensitivity (95% CI, 88%-97%), 96% specificity (95% CI, 94%-97%), and 73% positive predictive value (95% CI, 65%-80%), whereas the PE classifier achieved 94% sensitivity (95% CI, 89%-97%), 96% specificity (95% CI, 95%-97%), and 80% positive predictive value (95% CI, 73%-85%). CONCLUSIONS: Symbolic NLP can accurately identify VTEs from narrative radiology reports. This method could facilitate VTE surveillance and the evaluation of preventive measures.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".