SCREENING CONFOUND ESTABLISHED SCOPE-LIMITED svd metadata only
Scope caveat. This run is a pilot slice -- 1,853 of the 13,535 speakers measured elsewhere on svd. It may satisfy the repeat contract and still be a narrower measurement than a later run on the same corpus; both facts are shown because either alone would mislead.
| predictor | ROC-AUC | n |
|---|---|---|
| age alone | 0.8709 | 2,225 sessions |
| sex alone (negative control) | 0.5172 | 2,225 sessions |
age + sex, logistic, speaker-disjoint GroupKFold | 0.8768 | 2,225 sessions / 1,853 speaker groups |
Mean age 28.3 years healthy vs 51.0 pathological — a gap of 22.7 years over 1,356 pathological and 869 healthy sessions.
200 of 1,853 speakers contribute more than one session (max 24). A recording-level split leaks every one of them across folds.
Published benchmark for orientation: UAR 85.22 (arXiv:2410.10537) — different metric, indicative only
| artifact | autoresearch_results/F1_demographic_baseline.json · view raw · last written 2026-07-25 21:28 UTC |
| artifact_md5 | 2ee9852a19ede31c68107684b97bd308 |
| source | C:\Users\evija\auto-research-voice-based-disease-detection\data\raw\svd_meta\voice_data.csv |
| written up as | F1 in FINDINGS.md |
No 7-step reasoning entry exists for this run. The sibling convention renders Diagnosis · Citations · Hypothesis · Prediction · Verdict · Learning on every per-experiment page, sourced from a reasoning_annotations.json. This program never wrote one, and neither does it keep an append-only experiment_log.jsonl; the run history is the set of artifact files listed on the run ledger, ordered only by their write time. The pre-registered prediction for this run does exist — it is on the hypothesis registry, but this run predates the registry and has no pre-registered prediction of its own. Everything else on that convention’s checklist is absent here and is shown as absent rather than quietly dropped.