Machine-Learning Derived Relationships Between Sleep Variables and Mortality in Patients With Heart Failure and Cheyne-Stokes Respiration
Bibliographic record
Abstract
Abstract Background: In heart failure and reduced ejection fraction (HFrEF) patients, Cheyne-Stokes respiration (CSR) cycle length (CL) and lung-to-finger-circulation-time (LFCT) are directly related to circulation time and inversely proportional to cardiac output. It is therefore possible that CSR characteristics and LFCT may be related to mortality risk in patients with HFrEF. We hypothesized, in patients with CSR from the ADVENT-HF trial, that since longer CSR CL and LFCT indicate worse cardiac function, longer CSR cycling characteristics and LFCT are associated with higher mortality. Methods: In 195 patients with HFrEF (of whom 59 died over a maximum follow-up of 5 years), from 20 CSR cycles (10 in each of the first and last half of the night), we averaged CSR CL from the beginning of 1 apnea to the start of the next as well as apnea length (AL), hyperpnea length (HL), LFCT, respiratory rate, arterial oxygen saturation (SaO2), time to peak tidal volume (TPTV) during hyperpnea, and AHI (Figure 1). Clinical variables entered included sex, age, BMI, left ventricular ejection fraction (LVEF), atrial fibrillation, myocardial infarction, moderate-to-severe mitral regurgitation, New York Heart Association class, and peak-flow-adaptive-servo-ventilation treatment. A random forest feature selection technique was employed. The algorithm then built multiple decision trees, each considering different combinations of variables to predict mortality. The importance of a feature is determined by how much the mortality predictions worsen when the values of a feature are randomly mixed. A score above 0.05 indicates importance. Subsequently, we selected the most influential one-third of the features. We developed various classifiers using two distinct datasets: one based solely on sleep-related features and the other incorporating a broader range of variables, including demographic, cardiac, and sleep-related factors. The models were tested on 20% of unseen data for evaluation. Results: The most significant predictors of mortality, when using only sleep-related variables, included the average minimum (SaO2) (0.10), CL (0.09), LFCT (0.08), and AHI (0.07), resulting in a prediction accuracy of 71.8%. When incorporating cardiac and demographic variables, key predictors included TPTV (0.11), CL (0.10), LVEF (0.09), LFCT (0.09), age (0.09), and the average maximum SaO2 (0.08), leading to an improved prediction accuracy of 79.5%. Conclusion: These findings suggest that besides age and LVEF, CSR features related to cardiac output are significant predictors of mortality in HF patients with CSR. Our results underscore the potential significance of detailed polysomnographic analyses to assist in mortality risk stratification in this population.
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 | 0.001 | 0.004 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 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) | 0.001 | 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 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".