AI Science Tracker
EN

Machine learning links a three-gene lung-cancer score to retrospective survival differences

The Third People’s Hospital of Chengdu · 28 May 2026 · biology

Discovery

Machine learning links a three-gene lung-cancer score to retrospective survival differences

Source check: Source checked. Not independent scientific confirmation.

Finding: LASSO and random-forest selection identified a three-gene lung-adenocarcinoma score associated with observed survival in public cohorts. Comparison cohorts showed limited and inconsistent discrimination: TCGA five-year AUC was approximately 0.50, while another cohort reached approximately 0.68. This supports a bounded retrospective association, without established clinical utility or causal gene effects.

Technology: LASSO and random forests for prognostic gene selection; consensus clustering for RAS expression subtypes

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

AI contribution
Learned LASSO and random-forest feature selection produced the three-gene signature whose relationship with measured patient survival was evaluated across cohorts. General clustering benchmarks and untested drug predictions are not the admitted finding.
Test and result
The source reports 624 GEO lung-adenocarcinoma profiles used for derivation. LASSO with ten-fold cross-validation and random-forest feature selection intersected at MAPK10, PLA2G12B and SHC3. The resulting score was compared with observed survival in TCGA and GSE13213. Reported one-, three- and five-year AUCs were 0.608, 0.535 and 0.496 in TCGA, and 0.620, 0.654 and 0.682 in GSE13213.
Limitation / still unknown
Outcome-optimized risk cutoffs limit interpretation of group survival differences as locked external prediction. Exact cohort eligibility accounting and supplementary implementation were not reconstructed. Weak TCGA discrimination limits clinical claims; decision-curve calculations do not demonstrate actual patient benefit. Transcriptomic associations and drug predictions do not establish causal mechanisms or treatment efficacy. Original online date is May 28, 2026; the assigned December issue is later.
Evidence sources and annotation context

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

Evidence sources: https://link.springer.com/article/10.1007/s12672-026-05282-9

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: Read actual primary cohort acquisition, consensus clustering, gene selection, LASSO/random-forest and risk-score evaluation Methods, cohort survival/AUC Results and limitations. Supplementary code and individual-patient eligibility were not independently reconstructed. Exact original article front DOI/title and fresh PubMed primary relationship metadata checked..

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.

Open in the full catalog for filters, charts and related discoveries. Browse biology discoveries.