A Machine Learning Model for the Accurate Prediction of 1-Year Survival in TAVI Patients: A Retrospective Observational Cohort Study
Source check: Source checked. Not independent scientific confirmation.
Finding: An earlier PMU Science Get Together abstract reported ML prediction of one-year mortality in the same 629-patient TAVI cohort using the same model family and reject option (NPV 98%, PPV 89% after abstaining on 145/447); the later journal article reports revised metrics (NPV 96%, PPV 92%). The retrospective, single-center result remains limited by selective abstention. Machine-learning models predicted one-year survival after TAVI, with 10-fold cross-validation and a separate test sample; a random-forest reject-option subgroup yielded 96% accuracy. This retrospective cohort result is limited by selective abstention and single-center data.
Technology: Five ML approaches including RF/SVM trained in learning samples using 10-fold CV, then evaluated in held-out test data. 629 transfemoral TAVI patients; 565 had one-year outcome. With reject option, model predicts for a subset (~81% of original cohort) with reported accuracy 96%, NPV 96%, PPV 92%.
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- Five ML approaches including RF/SVM trained in learning samples using 10-fold CV, then evaluated in held-out test data. 629 transfemoral TAVI patients; 565 had one-year outcome. With reject option, model predicts for a subset (~81% of original cohort) with reported accuracy 96%, NPV 96%, PPV 92%.
- Test and result
- Five ML approaches including RF/SVM trained in learning samples using 10-fold CV, then evaluated in held-out test data. 629 transfemoral TAVI patients; 565 had one-year outcome. With reject option, model predicts for a subset (~81% of original cohort) with reported accuracy 96%, NPV 96%, PPV 92%.
- Limitation / still unknown
- Machine-learning models predicted one-year survival after TAVI, with 10-fold cross-validation and a separate test sample; a random-forest reject-option subgroup yielded 96% accuracy. This retrospective cohort result is limited by selective abstention and single-center data. The source does not establish benefits beyond the reported evaluation.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://mdpi-res.com/d_attachment/jcm/jcm-12-05481/article_deploy/jcm-12-05481.pdf · https://www.pmu.ac.at/fileadmin/CONTENT/BEREICHE/Forschungsmanagement/Dokumente/FM_Dokumente/SGT/Abstractband___Wissenschaftsbericht/SGT23_Abstractband_Digital.pdf · https://www.pmu.ac.at/en/research-innovation/research-management/news-events/science-get-together-sgt/vsgt23 · https://api.crossref.org/works/10.3390/jcm12175481
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: other. 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.