Gradient-boosted trees distinguish historical autism diagnostic categories
Source check: Source checked. Not independent scientific confirmation.
Finding: Gradient-boosted trees classified historical DSM-IV autism categories using assessment scores, demographics and comorbidities. In a person-separated SPARK test set, the multiclass model correctly classified 80.5% of individuals; 12.6% were assigned a different autism-spectrum category. This retrospective result does not establish screening performance under current DSM-5 criteria.
Technology: Gradient-boosted decision trees
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- Gradient-boosted trees learned associations between clinical/assessment features and recorded diagnostic categories, then classified unseen individuals from the same retrospective dataset.
- Test and result
- SPARK supplied 36,965 individuals, split 80:20 without person overlap; fivefold training-only grid search optimized gradient-boosted trees. The multiclass model correctly classified 5,951 of 7,393 test individuals (80.5%), with 933 assigned a different autism-spectrum category. Separate one-versus-reference models yielded AUROCs 0.863–0.980. ABIDE supported a separately developed model, not external testing of the SPARK model.
- Limitation / still unknown
- Historical professional DSM-IV diagnoses were the reference; absence of diagnostic codes defined non-spectrum controls. Assessment inputs overlap with information used in diagnostic evaluation, limiting independent screening interpretation. SPARK and ABIDE models used different inputs and were fitted separately. No prospective workflow, patient benefit or current DSM-5 diagnostic validation was established; PDD-NOS positive predictive value was low.
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/PMC11461775/
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 dataset/reference definition, person-separated partition, gradient-boosted model/tuning Methods, input table, classification Results and limitations. Supplementary ABIDE tables were not independently inspected; their outcomes are not represented as external validation. Exact primary identity/date and fresh primary PubMed correction/retraction relationships 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 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.