Development of a Clinical Implementation Framework for LIRIC (Liver Cancer Risk Computation) Methods

Hepatocellular carcinoma (HCC) is commonly diagnosed as advanced disease, with a low overall survival rate. However, patients with liver cirrhosis that undergo routine screening have significantly higher rates of early-stage cancer detection and improved survival rates. Despite this, screening is severely underutilized by eligible patients. Moreover, screening eligibility criteria do not include those cases of HCC arising in patients without underlying liver cirrhosis. 

To address this need, we previously developed, validated, and simulated the deployment of LIRIC (LIver cancer RIsk Computation) models. LIRIC models can identify individuals at high-risk for HCC, to both improve HCC diagnostic accuracy and prevent missed diagnoses. LIRIC models were developed on routinely collected Electronic Health Record (EHR) data from multiple institutions using a federated EHR network.

The objective of this proposal is to develop the framework for clinical implementation of LIRIC (LIver cancer RIsk Computation) models, through the following aims: improving the interpretability of LIRIC models, establishing decision rules, and prospective model validation.

Our goal is to produce generalizable, scalable, automated, and interpretable models that can be implemented in the clinic for personalized HCC surveillance, allowing increased rates of early cancer detection and curative intervention, and decreasing costs to the healthcare system