AI Science Tracker
EN

Learned feature selection identifies exploratory lymphedema-dose associations

Northwestern University; University of Pennsylvania · 28 Feb 2024 · medicine

Discovery

Learned feature selection identifies exploratory lymphedema-dose associations

Source check: Source checked. Not independent scientific confirmation.

Finding: In76 treated head-and-neck cancer patients, an ensemble feature-selection analysis linked selected lymphatic-region dose features with recorded lymphedema. Subsequent competing-risk models described associations with external and internal swelling. These retrospective results do not establish a deployable forecast or causal dose effect.

Technology: Support vector machine recursive feature elimination; random forest; XGBoost; LASSO

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

AI contribution
Learned classifiers rank and eliminate features, enabling exploratory dose-region association analysis.
Test and result
Learned feature ranking reduced clinicopathologic and dosimetric inputs before same-cohort competing-risk association analyses.
Limitation / still unknown
Selection and cutoff derivation reused the small cohort. Follow-up duration appeared among classification predictors, precluding an admission-time prediction interpretation. External/internal describe edema location, not independent validation cohorts. Prior cohort reporting, incomplete nesting and unmeasured confounding limit novelty and inference.
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/PMC10915511/

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: Actual primary cohort/endpoint, feature-selection/model and competing-risk Methods/Results/limitations; exact front/current notice metadata..

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