Learned screening identifies liver-expression differences across archived cohorts
Source check: Source checked. Not independent scientific confirmation.
Finding: Three feature-selection methods nominated five expression candidates for fatty liver disease. The source reports concordant expression differences in pooled external liver datasets, with individual-marker AUCs of 0.652–0.751. This is a retrospective tissue association, not a diagnostic test or a causal immune mechanism.
Technology: LASSO; support-vector-machine recursive feature elimination; random forest
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- LASSO,SVM-RFE and random forest fitted disease-label feature selectors; their intersection selected BBOX1,FOSB,NR4A2,RAB26 and SOCS2 for later expression comparisons.
- Test and result
- Training combined50 disease and31 control samples. External pooled datasets contained271 disease and85 control samples; source-reported marker AUCs were0.734,0.751,0.709,0.670 and0.652.
- Limitation / still unknown
- Original datasets and batch correction can confound retrospective comparisons; no prospective or separate-team clinical validation is established. BBOX1 was not significantly different in the206-case/10-control dataset alone, despite significance in the pooled set. Source figure captions inconsistently call the five-marker intersection seven markers; RF selection counts also vary. Earlier roles of FOSB and SOCS2 are acknowledged. Protein experiments, mechanistic claims and inferred immune-cell causality are outside this represented result.
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/PMC11301246/
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: Primary front;computational Methods2.1,2.5 and Results3.4–3.5;primary deposited metadata status;experimental procedures not reviewed.
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 1/5 · 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.