Machine Learning for Prediction of Therapy Response in Autoimmune Hepatitis

Machine Learning for Prediction of Therapy Response in Autoimmune Hepatitis

The 5th International Autoimmune Hepatitis Group (IAIHG) research workshop emphasized the integration of large clinical cohorts with artificial intelligence (AI) for enhanced prediction of therapy responses and outcomes in Autoimmune Hepatitis (AIH). This project aims to develop and validate machine learning (ML) models using data from the R-Liver registry and other international cohorts. After rigorous preprocessing to ensure data uniformity and quality, the investigators will identify and characterize factors influencing therapy response. They will then implement ML models to predict complete biochemical response (CBR) at 6 and 12 months, using five-fold cross-validation, and validate these models in external cohorts from Spain, Canada, and the international AIH group, ensuring robustness and generalizability.

Finally, the investigators will prospectively validate the models in newly registered cases, assessing both short-term and long-term outcomes. This project seeks to advance personalized treatment strategies in AIH, facilitating timely adjustments in therapy and improving patient prognosis through AI-driven decision support. This projects' interdisciplinary team, with expertise in clinical AI and hepatology, is well-equipped to address these challenges and enhance the clinical management of AIH.

Inclusion Criteria:

• Diagnosis of Autoimmune hepatitis
• Consent to provide data to the specific registry
• available follow-up data at least within the first 12 months

Exclusion Criteria:

• variant syndromes
• extrahepatic autoimmune disease
• immunosuppressive treatment at diagnosis