AI Science Tracker
EN

Southern Illinois University

1 discovery from 2026

Discoveries

Acquisition year matched breast-thermography model discrimination

Source check: Source checked. Not independent scientific confirmation.

Finding: On a held-out set of 52 patients, acquisition year alone separated normal and sick labels with an AUROC of 0.974, within the range of the trained imaging models. On a separate 24-patient test set, confidence intervals for benign-versus-malignant classification all included chance.

Technology: BiomedCLIP, general CLIP variants and ViT-B/16; federated training simulation

Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
The learned thermography classifiers supplied the scores audited against acquisition dates and evaluated on the pathology distinction.
Test and result
The authors separated patients across splits, tested acquisition year as a metadata-only score and compared it with the image-model outputs.
Limitation / still unknown
Only two acquisition years included both normal and sick labels. These data cannot separate acquisition effects from pathology, and the small malignancy test does not establish clinical utility.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://arxiv.org/abs/2610.36379v1 · https://arxiv.org/html/2610.36379v1

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 21 Nov 2026. Review scope: §§3–4.4, patient partitions, Tables 1–2, acquisition-period control.

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open the interactive profile for the timeline, or browse all organizations and model families.