Abstract 14935: Deep Learning Pipeline for Frailty Screening Using the 12-Lead Electrocardiogram
Bibliographic record
Abstract
Introduction: The electrocardiogram (ECG) contains information about age-related changes in cardiovascular physiology, which have been linked with the frailty syndrome. Hypothesis: We sought to develop and validate a predictive model leveraging the 12-lead ECG to screen for frailty as defined by a prospective reference standard. Methods: We conducted a population-based cohort study using data from the Canadian Longitudinal Study of Aging (CLSA). From 2010-2015, the CLSA enlisted a diverse and multi-ethnic sample of community-dwelling adults 45-85 years of age. Comprehensive phenotyping was performed through interviews at participants’ homes and assessments at data collection sites. Frailty was quantified by the 110-item Frailty Index (FI) Composite, consisting of self-reported comorbidities, blood tests, physical performance tests, body composition tests, cardiovascular and pulmonary tests, cognitive and sensory tests. After dividing our sample into training (80%) and test (20%) sets, we developed an end-to-end deep neural network to predict the FI score based on the 12-lead ECG time series. Results: A total of 26,700 ECGs with paired FI scores were evaluated. For classification of FI quintiles, a bidirectional long short-term memory (BiLSTM) neural network with a cross-entropy loss function achieved a 5-fold mean area under the receiver operating characteristics curve (AUROC) of 0.70 and area under the precision-recall curve (AUPRC) of 0.36. Predictive performance was superior for classification of the first (most robust) quintile that had AUROC 0.79 and AUPRC 0.46, and the fifth (most frail) quintile that had AUROC 0.79 and AUPRC 0.56, as compared to the middle quintiles that had AUROC 0.60-0.69 and AUPRC 0.27-0.29. Conclusions: Our deep learning model can be used to screen for high or low levels of frailty based on the readily available 12-lead ECG. Additional research is underway to gain insights into other representations of the ECG signal and relative importance of the ECG features.
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 | 0.002 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.007 | 0.004 |
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".