AI Science Tracker
EN

Auburn University

1 discovery from 2026

Discoveries

Clinical model distinguishes selected rare-infection diagnoses in archived admissions

Source check: Source checked. Not independent scientific confirmation.

Finding: HistGradientBoosting used early-admission clinical records to distinguish a selected composite of rare infections from urgent-admission controls. The primary holdout table reports an ROC area of 0.937; a separate comparative validation table reports mean sensitivity of 61.9%. This is internal retrospective evaluation, without demonstrated earlier diagnosis or clinical benefit.

Technology: Histogram gradient boosting; SHAP

Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
A fitted boosted-tree classifier maps early clinical and laboratory features to chart-confirmed diagnostic labels; SHAP describes model-used features.
Test and result
The study used1494 cases and10000 sampled controls from2010–2023, with a70:30 patient split. Training-only cross-validation, SMOTE and isotonic recalibration are described. The paper distinguishes a448-case/3001-control test composition.
Limitation / still unknown
The abstract and cross-validation table report AUC0.947 while the separate holdout/ablation table reports0.937; these are retained as different reported evaluations, not silently equated. Early predictor timing does not prove predictions preceded actual diagnosis. The selected six-disease composite and sampled-control prevalence limit calibration and clinical transfer. Supplementary inclusion details were not read. No pathogen methods, treatment protocol, real-world alert deployment or prospective outcome benefit is represented. The 61.9% sensitivity comes from the comparative validation table, not an authenticated operating point for the separate holdout AUC of 0.937.
Evidence sources and annotation context

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

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC13682164/ · https://doi.org/10.1016/j.ijmedinf.2026.106688

Source checked · editorial details

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

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 21 Nov 2026. Review scope: Original primary clinical/computational Methods, model evaluation and high-level Results read; no operational biological protocols.

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.