Identify Patients with Congestive Heart Failure through Analyzing Free-Text Clinical Notes
Bibliographic record
Abstract
IntroductionA number of challenges exist in analyzing unstructured free text data in electronic medical records (EMRs). EMR text are difficult to represent and model due to their high dimensionality, heterogeneity, sparsity, incompleteness, random errors and the presence of noise.
 Objectives and ApproachStandard Natural Language Processing (NLP) tools make errors when applied to clinical notes due to physician use of unconventional language, involving polysemy, abbreviations, ambiguity, misspelling, variations, and negation.
 This paper presents a novel NLP framework, “Clinical Learning On Natural Expression” (CLONE), to automatically learn from a large primary care EMR database, analyzing free text clinical notes from primary care practices. CLONE’s predictive clinical models using text mining and neural network approach to extract features to identify patterns. To demonstrate effectiveness, we evaluate CLONE’s ability in a case study to identify patients with a specific chronic condition: congestive heart failure (CHF).
 ResultsA random selected sample of 7500 patients from Electronic Medical Record Administrative data Linked Database (EMRALD) is used. In this dataset, each patient’s medical chart includes a reference standard, manually reviewed by medical practitioners. Prevalence of CHF is approximately 2%. The low prevalence leads to another challenging problem in machine learning: imbalanced datasets. After pre-processing, we build deep learning models to represent and extract important medical information from free text to identify CHF patients through analyzing patient charts. We evaluated the effectiveness of CLONE by comparing the predicted labels with the standard references on a holdout test dataset. Comparing it with a number of alternative algorithms, we improve the overall accuracy to over 90% on a test dataset.
 Conclusion/ImplicationsAs the role of NLP in EMR data expands, the CLONE natural language processing framework can lead to substantial reduction in manual processing, while improving predictive accuracy.
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 | 0.002 | 0.005 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.007 |
| Open science | 0.006 | 0.001 |
| 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".