Experiment workflow
Cardiology follows a left-to-right research flow across four top-level tabs. You can revisit earlier steps at any time; downstream cohort views depend on completed HRV runs for linked subjects.
Create study → Data (upload, subjects) → Analyze (pipelines, cohort) → Interpret → History
Opening a study (/experiments/{id}) takes you straight to the Data tab — there is no separate study-overview landing page.
Data — bring data in, check readiness
The Data tab has two sub-tabs:
- Datasets — paginated list of every uploaded recording for this study, plus Upload data and Manage subjects actions.
- Readiness — a subject-by-subject matrix showing modality availability (ECG/RR/PPG/BP/respiration), pipeline completion (HRV/PRV/BP analysis/arrhythmia/RSA-baroreflex), outcomes present, motion quality checked, and unresolved failed runs. Use this to answer "what can I analyze right now?" across the whole cohort.
See Data upload.
Analyze — run pipelines and compare cohorts
The Analyze tab has four sub-tabs, in the order you'd typically use them:
- Signal Quality — per-dataset waveform SQI (ECG/PPG) or interval-based quality (RR), plus Preprocess all pending to backfill quality for every dataset that hasn't been preprocessed yet.
- Signal Explorer — pick one dataset and open its full inspector: Explorer (waveform/tachogram), Overview (run pipelines, view metrics), Annotations, and Runs (history, compare runs).
- Bulk Pipelines — run HRV, PRV, BP analysis, and arrhythmia burden across every eligible dataset in the study at once, instead of one at a time in Signal Explorer. Use this after Signal Quality to catch up any datasets still missing a metric, before comparing cohorts.
- Cohort comparison — filter subjects, batch-fill missing HRV, compare groups with statistical tests, review outliers, and compute research-only risk scores.
See Dataset inspector and Subjects and cohorts.
Interpret — study-level synthesis
The Interpret tab generates one grounded interpretation across every completed analysis run in the study — HRV, BP, arrhythmia, cohort comparisons, whatever exists — with cited metrics, key findings, and hypotheses. Click Generate interpretation (or Regenerate once one exists). This is separate from the per-run/per-cohort/per-subject Interpret buttons described in AI interpretation, which scope to one run or one cohort comparison instead of the whole study.
History — every run
A flat list of every pipeline run for the study, with status, completion time, and a delete action.
Typical paths
| Study type | Path |
|---|---|
| Single Holter review | Upload → Signal Explorer: preprocess → HRV → explorer → interpret |
| RR wearable cohort | Import subjects → upload RR CSVs → Bulk Pipelines (or Cohort comparison's batch panel) → cohort compare |
| Multimodal subject | Link RR + BP → HRV + analyze BP → baroreflex → mechanistic interpret |