Cardia-AI: Passive Cardiac Event Monitoring Using Smartwatch Sensors and Predictive Analysis via Large Language Models
Bibliographic record
Abstract
Cardiovascular diseases require continuous, context-aware monitoring, i.e., combining day-to-day wearable signals with recent diagnoses, medications, and symptom reports, rather than isolated clinic visits or single-spot measurements. We developed Cardia-AI, a proof-of-concept pipeline that time-aligns smartwatch signals (heart rate, blood pressure, oxygen saturation) with a patient's longitudinal electronic health record (EHR) and uses a compact medical language model with retrieval to produce grounded educational summaries. Cardia-AI assembles time-aligned summaries through a select, compile, and ask pipeline, and uses a lightweight medical large language model (BioMistral-7B) with retrieval from curated sources. Guardrails constrain outputs to education and navigation, and the system incorporates explicit escalation guidance when red-flag symptoms are present in the prompt context. In two scenario-based validations (cardiometabolic education; early post-angioplasty recovery), Cardia-AI compiled synchronized smartwatch trends with EHR entries, referenced the exact measurements and diagnoses present in the prompt, and recorded transcripts for audit and reproducibility; no outcomes or accuracy endpoints were assessed. We did not evaluate clinical effectiveness or diagnostic accuracy; no patient outcomes were measured. This work reports feasibility and safety guardrails only, with prospective evaluations planned. These demonstrations suggest that pairing wearable streams with a compact, domain-tuned language model may lower cognitive load from multi-panel charts and shorten the path from symptom onset to appropriate follow-up under clinician oversight.
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 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| 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".