Environmental exposures classify recorded diabetes in NHANES
Source check: Source checked. Not independent scientific confirmation.
Finding: A learned exposure model distinguished recorded diabetes in a random NHANES holdout with AUC .78. The analysis used 8,501 participants from 2005–2016. It establishes a retrospective predictive association, with no proof that the measured chemicals cause diabetes or predict its future onset.
Technology: Random-forest feature screening and LASSO regression
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- Learned feature selection and penalized regression quantify the predictive association between measured environmental exposures and recorded diabetes.
- Test and result
- The original abstract reports 5,953 discovery and 2,548 validation participants, training-set ten-fold feature selection using RF and LASSO, and diabetes AUC .80 versus .78. Separate linear-model validation R² values were .40 for insulin, .15 for fasting glucose and .17 for two-hour glucose; these are not additional independent discovery records.
- Limitation / still unknown
- This review had the complete original abstract and metadata, not full computational Methods. Exact feature identities, endpoint definition, sampling/measurement subsets, survey weighting, missingness and detailed preprocessing containment were not authenticated. The random validation set comes from the same survey periods; no external cohort or prospective incident-diabetes prediction is established. Chemical levels can reflect correlated exposure or disease-related processes; prediction does not establish causation. A bibliographic trial label does not make this observational analysis a randomized intervention.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pubmed.ncbi.nlm.nih.gov/34597539/
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: Complete original primary abstract, dates and institutional affiliations actually read; full-text details were not available in this bounded review..
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.