Integrating Heterogeneous Echocardiography AI Outputs in Language Space: Learnable Late Fusion with a Supervised LLM
Source check: Source checked. Not independent scientific confirmation.
Finding: Patient-level 3-fold OOF results are comparable to best report-rule/LightGBM strategies and outperform an untuned general-purpose LLM; no material superiority over best baseline is claimed. This is the result-bearing preprint version dated 2026-10-29; peer-review status is not established by the reviewed source.
Technology: Supervised LLM combines view-specific finding statements and measurements for four echocardiographic targets; retrospective single-center cohort is 859 exams from 646 patients.
Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- Supervised LLM combines view-specific finding statements and measurements for four echocardiographic targets; retrospective single-center cohort is 859 exams from 646 patients.
- Test and result
- Patient-level 3-fold OOF results are comparable to best report-rule/LightGBM strategies and outperform an untuned general-purpose LLM; no material superiority over best baseline is claimed. Patient-level splits and out-of-fold predictions validate within this cohort; source explicitly limits generalization because no external institution/future-period cohort was tested.
- Limitation / still unknown
- nostic-validation result, not evidence of clinical benefit or external generalization.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://arxiv.org/pdf/2610.39464
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: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..
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.