Deep Learning and NLP For Knowledge Extraction from Laboratory Reports
Bibliographic record
Abstract
IntroductionDue to the ever-growing volume and complexity of clinical data, it has become a tedious task to extract information from data for secondary uses such as decision support, quality assurance, and outcome analysis. Recently, there have been great advances in Natural Language Processing (NLP) approaches that automate knowledge extraction from clinical reports in order to save costs and improve efficiency. Objectives/ApproachOur goal is the development of an NLP tool designed to automatically extract and encode clinical information from laboratory reports. This study describes and evaluates our NLP tool on provincial repositories of laboratory tests and results called Ontario Laboratory Information System (OLIS). OLIS is an electronic system that covers >200 labs and stores patients’ current and past test results as patients move through different areas of the healthcare system. Our NLP tool is a modular system of pipelined components including Named Entity Recognition module for extracting mentions of virus and test mentions and inference to combine extracted entities into a meaningful outcome. ResultsInitial analyses were conducted on a segment of OLIS related to laboratory tests for respiratory viruses. This data included over a million observations corresponding to ~100 Logical Observation Identifiers Names and Codes (LOINC), with >40,000 unique strings. The clinical text was cleaned, tokenized, and parsed using an in-house text algorithm that was continually refined with manual review from clinical experts. This data was then encoded as virus and test types to be used as a ground truth. The NLP tool was built on ground truth data and achieved an accuracy greater than 95%. Conclusion/ImplicationsApproaches like these can be applied to many areas of health research that make use of clinical reports. Our methods, when optimized and validated, can be deployed into clinical systems to provide on-the-spot analysis of various laboratory reports.
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 | 0.000 | 0.003 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 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".