EUVIMEDEUVIMED entdecken
Crossref

Prediction Beyond Established Models in Steatotic Liver Disease: Which New Predictors Add Value?

Ali Zifan, Panos Liatsis

Liver International · 2026-12 · 46 (12)

DOI: 10.1111/liv.70952 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Liver International, online erschienen am 18.11.2026. Verfasst von Ali Zifan und 1 weiteren Autor:innen. Die Arbeit nennt 49 Literaturangaben. Gefördert von: Khalifa University of Science, Technology and Research.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background & Aims Biomarkers, non‐invasive tests (NITs), imaging, genetics and machine learning are increasingly used for risk stratification in metabolic dysfunction‐associated steatotic liver disease (MASLD) and related liver disease, yet their incremental predictive value beyond established models remains unclear. We developed an evidence map to characterise incremental predictive value beyond established clinical models.

Methods

We searched PubMed using four search families, supplemented by ClinicalTrials.gov and seed‐literature/backward‐citation searching. Searches covered January 2000 through August 26, 2026 without a language restriction. Records were normalised, deduplicated and processed through deterministic prescreening followed by a downstream targeted‐selection stage; the archived workflow does not preserve the exact final targeting rule used to define the structured analytic corpus. A locally hosted large language model (LLM) pipeline extracted prespecified evidence fields, with deterministic consistency checks and study‐level adjudication of all 44 full‐text candidates. A post hoc blinded audit of 100 analytic records assessed agreement between the automated candidate flag and blinded reference adjudication.

Results

Of 3230 unique discovery records, 1352 entered structured analysis and 44 underwent full‐text adjudication. Twenty‐two (50.0%) demonstrated incremental predictive value, including 15 without reported validation, 5 with internal validation and 2 with independent external or temporal validation. The latter represented 2/44 (4.5%) of strict full‐text candidates and 2/1352 (0.15%) of the structured analytic corpus; the 0.15% figure is an evidence‐funnel yield, not a prevalence estimate. The two independently validated studies addressed distinct contexts: PNPLA3/TM6SF2 augmentation of established non‐invasive fibrosis tests for advanced fibrosis in MASLD with type 2 diabetes, and the Non‐Invasive CSPH Estimated Risk (NICER) model incorporating spleen stiffness measurement (SSM) for clinically significant portal hypertension (CSPH) in compensated advanced chronic liver disease. Calibration was explicitly identifiable in 6 of 22 incremental studies; 5 were judged adequate and 1 unclear.

Conclusions

The literature contains substantial incremental model development but limited independently validated evidence of added predictive value beyond established frameworks. The two validated studies should be interpreted within their distinct populations, endpoints and clinical pathways.

Förderung

  • Khalifa University of Science, Technology and Research

Zitieren

APA

Zifan, A., & Liatsis, P. (2026). Prediction Beyond Established Models in Steatotic Liver Disease: Which New Predictors Add Value?. Liver International, 46(12). https://doi.org/10.1111/liv.70952

Vancouver

Zifan A, Liatsis P. Prediction Beyond Established Models in Steatotic Liver Disease: Which New Predictors Add Value?. Liver International. 2026;46(12). doi:10.1111/liv.70952

BibTeX
@article{Zifan2026,
  author = {Zifan, Ali and Liatsis, Panos},
  title = {Prediction Beyond Established Models in Steatotic Liver Disease: Which New Predictors Add Value?},
  journal = {Liver International},
  year = {2026},
  volume = {46},
  number = {12},
  doi = {10.1111/liv.70952}
}
Kostenlos starten →