AI Models vs Non-Invasive Fibrosis Scores in MAFLD Diagnosis

Assessing the Utility of AI Models in MAFLD Diagnosis: Comparison With Traditional Non-Invasive Fibrosis Scores.

Status

Completed

Phases

Unknown

Study type

Observational

Source

ClinicalTrials.gov

Registry ID

NCT07305636

Acronym

MAFLD-AI

Enrollment

522

Registered

2025-12-26

Start date

2025-05-13

Completion date

2025-11-30

Last updated

2025-12-26

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

AI (Artificial Intelligence), MAFLD

Brief summary

This study evaluates the accuracy of artificial intelligence (AI) models using FibroScan and clinical data to predict hepatic fibrosis in Egyptian patients with metabolic-associated fatty liver disease (MAFLD). The performance of the AI models will be compared with conventional noninvasive fibrosis scores (FIB-4, APRI, NAFLD fibrosis score, and FAST). The goal is to improve early, noninvasive diagnosis of fibrosis and reduce reliance on liver biopsy.

Interventions

None listed

Study design

Observational model

OTHER

Time perspective

CROSS_SECTIONAL

Eligibility

Sex/Gender

ALL

Age

18 Days to No maximum

Inclusion criteria

\- Adults ≥18 years. Diagnosed with MAFLD according to international criteria (hepatic steatosis with metabolic dysfunction). Valid FibroScan evaluation with available LSM and CAP values.

Exclusion criteria

* Excessive alcohol intake (\>30 g/day for men, \>20 g/day for women). Chronic viral hepatitis (HBV or HCV). Autoimmune hepatitis. Known malignancy. Pregnancy. Refusal to participate.

Design outcomes

Primary

Measure	Time frame
Measure diagnostic accuracy of AI models in predicting hepatic fibrosis stage (F0-F4)	At enrollment (single cross-sectional assessment).

Countries

Egypt

Outcome results