Automated processing of electronic medical records is a reliablemethod of determining aspirin use in populations at risk forcardiovascular events
Bibliographic record
Abstract
BACKGROUND: Low-dose aspirin reduces cardiovascular risk; however, monitoring over-the-counter medication use relies on the time-consuming and costly manual review of medical records. Our objective is to validate natural language processing (NLP) of the electronic medical record (EMR) for extracting medication exposure and contraindication information. METHODS: The text of EMRs for 499 patients with type 2 diabetes was searched using NLP for evidence of aspirin use and its contraindications. The results were compared to a standardised manual records review. RESULTS: Of the 499 patients, 351 (70%) were using aspirin and 148 (30%) were not, according to manual review. NLP correctly identified 346 of the 351 aspirin-positive and 134 of the 148 aspirin-negative patients, indicating a sensitivity of 99% (95% CI 97-100) and specificity of 91% (95% CI 88-97). Of the 148 aspirin-negative patients, 66 (45%) had contraindications and 82 (55%) did not, according to manual review. NLP search for contraindications correctly identified 61 of the 66 patients with contraindications and 58 of the 82 patients without, yielding a sensitivity of 92% (95% CI 84-97) and a specificity of 71% (95% CI 60-80). CONCLUSIONS: NLP of the EMR is accurate in ascertaining documented aspirin use and could potentially be used for epidemiological research as a source of cardiovascular risk factor information.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.033 | 0.152 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.012 | 0.007 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.003 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.002 |
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 source (direct Gemma or distilled Codex), 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".