A serum-protein classifier fails to reproduce its early-infection discrimination in a second clinical cohort
Source check: Source checked. Not independent scientific confirmation.
Finding: A random-forest model using four circulating proteins distinguished later invasive-aspergillosis cases from non-infected hematology controls in its derivation data, but the source reports that it failed to reproduce this performance in a second cohort. This negative retrospective validation is separate from the paper’s conventional differential-protein findings and does not establish a usable early diagnostic test.
Technology: Random forest classification of circulating-protein measurements
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- A learned RF combined the four selected protein measurements into a case-classification rule. Its failure to transfer to a second observed clinical cohort is the represented clinical validation result; the separate IL-17C association was obtained through conventional differential analysis and is not attributed to AI.
- Test and result
- The discovery study included 33 cases and two matched 33-person control groups. The RF used four selected proteins measured more than ten days before clinical diagnosis, and was then assessed in a separate cohort nominally comprising 20 cases and 20 non-infected controls. The source reports 87% derivation accuracy and 47% in the second cohort, explicitly concluding that the model did not replicate.
- Limitation / still unknown
- The main paper does not fully specify RF tuning, the person-versus-serum-sample denominator for these accuracy figures, uncertainty intervals or all handling of repeated longitudinal samples. Cohorts came from the same center and differed in immune-status/treatment characteristics. No significant differential proteins distinguished cases from infected controls after multiplicity correction. The failed classifier is not evidence of an infection-specific mechanism, actionable screening rule or patient benefit.
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/s11046-024-00831-8
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 original clinical cohort/control and time-window definitions, RF construction and second-cohort model Results, conventional-marker analysis distinction and cohort/reference limitations. No biological operational protocol is represented or required. Original publisher February26 date verified. 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.