Generative AI in hepatology: Transforming multimodal patient-generated data into actionable insights

Cirrhosis care is inherently complex, marked by a high risk of acute decompensation and significant morbidity and mortality. Traditional episodic care models provide static snapshots of a patient's condition, limiting the ability to address dynamic changes in clinical status. Emerging at-home monitoring technologies and wearable devices present numerous opportunities to generate continuous clinical data, but the integration of this data into clinical workflows remains challenging. Large language models (LLMs) and generative AI (GenAI) technologies offer innovative solutions by enabling the acquisition, summarization, and actionable analysis of multimodal data generated by patients. This review explores the application of home-monitoring technologies for key complications of cirrhosis, including HE, ascites, and frailty. GenAI will enable the integration of these home-based data with canonical clinical data acquisition. We discuss the role of GenAI technologies and LLMs in processing multimodal data, supporting clinical decision-making, and creating autonomous artificial intelligence (AI) agents capable of triaging and summarizing patient data. In addition, we offer perspectives on the clinical evaluation of these emerging technologies. Finally, we close with a "not-so-distant" vision of GenAI-enabled at-home cirrhosis care.