Abstract 4139307: Machine Learning Identifies Predictors of Poor Outcomes in Patients with Heart Failure Presenting to the Emergency Department for Chest Pain
Bibliographic record
Abstract
Introduction: Heart failure (HF) is associated with unique comorbidities and sequelae, which can affect clinical presentation and patient outcomes. This is specifically challenging when patients are evaluated for suspected acute coronary syndrome (ACS). We sought to compare the most important predictors of poor outcomes in patients with and without HF seen in the emergency department (ED) for ACS. Methods: This was a secondary analysis of a prospective observational cohort study of consecutive patients seen for symptoms suggestive of ACS, such as chest pain (CP) and dyspnea, in the EDs of three UPMC-affiliated tertiary care hospitals (NCT04237688, clinicaltrials.gov). Primary outcome was 30-day major adverse cardiac events (MACE), adjudicated by two independent reviewers. Clinical data were collected form charts and we used KNN to impute missing data for features, most of which had less than 12.5% missingness. For features with greater than 12.5% missingness (i.e., BNP, Mg), binary indicators were added to flag missing values. Data were normalized using the Euclidean norm. Two random forest (RF) classifiers were trained using 10-fold cross validation with 71 manually selected features available early in the ED course (i.e., vital signs, labs, past medical history, ECG), and tested on patients with and without known HF. Model performance was evaluated using AUROC, and top features were identified with SHAP values. Results: The sample included 2400 patients (age 59 ± 16 years; 47% female, 41% Black, 15.9% ACS), of whom 438 had HF (age 66 ± 14 years; 45% female, 49% Black, 15.1% ACS). Individuals with HF were more likely to experience MACE (38% vs 23%, p <.001). The model had higher classification performance in patients without HF (AUC 0.89 vs 0.70). Figure 1 shows top MACE predictors for patients with and without HF. Conclusion: The RF model performed sub-optimally among patients with HF, who were more likely to have poor outcomes moderated by characteristics such as low pulse oximetry, BNP measurement, and anemia, which are related to pathophysiology of HF. Current routine ED assessment tools for CP (i.e., HEART score) do not translate well to patients with HF, which requires more elaborate diagnostic testing.
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 | 0.000 |
| 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".