How artificial intelligence and machine learning are changing the diagnosis and treatment of liver disease

AI-driven liver disease diagnosis
Machine learning for treatment planning
Predict disease progression
The future of hepatology
refer to
further reading

---

Artificial intelligence (AI) is an umbrella term that encompasses all computational processes designed to imitate and extend human intelligence for problem solving and decision-making. It is based on an algorithm or array of mathematical formulas that make up a specific computational learning method. Machine learning (ML) and deep learning (DL) use algorithms in more sophisticated ways to predict learned and new outcomes.

Diagnosis and Treatment of Liver Disease》 />

Hepatology relies heavily on imaging, which is an area where artificial intelligence can take full advantage. Machine learning is at work to extract rich information from imaging and clinical data to aid in non-invasive and accurate diagnosis of a variety of liver diseases.

AI can also help develop objective risk stratification scores, predict the course or treatment outcome of chronic lung disease or liver cancer, facilitate easier and more successful liver transplants, and develop hepatology quality indicators.

AI-driven liver disease diagnosis

Liver biopsy is the gold standard for many chronic liver diseases (CLD), including post-chronic hepatitis liver fibrosis, non-alcoholic fatty liver disease (NAFLD), cholangitis, and liver tumors or cysts.

Its invasiveness, particularly in patients already at clinically high risk for bleeding complications, and its complexity and cost hinder its use as a routine screening technique. Likewise, ultrasound imaging can be used to detect and classify many liver diseases, but requires a skilled operator and interpretation.

In this context, artificial intelligence in hepatology provides non-invasive diagnosis and extends expert skills to multiple medical centers. Machine learning and deep learning use algorithms that reflect or contribute to the development of currently accepted diagnostic guidelines and are refined through application to large imaging and clinical data sets.

Radiomics refers to the computerized processing of imaging data to generate and classify large amounts of information to produce predictions. However, radiomics is so emerging that standardized protocols and standards are still awaited.

AI in liver disease is exemplified by the radiomics fibrosis index (RFI). This new deep learning-based model uses enhanced magnetic resonance imaging (MRI) to predict the stage of liver fibrosis. It exceeds the performance of other non-invasive clinical tests, such as AST:platelet ratio and fibrosis-4 (FIB-4) index, obviating liver biopsy in many cases.

For example, liver fibrosis and chronic liver disease are common sequelae of chronic hepatitis B and chronic hepatitis C. Both outcomes also pose independent risks for hepatocellular carcinoma (HCC). Diagnostic liver biopsy is impractical to screen all patients with chronic hepatitis, but machine learning-based screening offers rich promise.

Likewise, NAFLD is one of the leading causes of cirrhosis and nonalcoholic steatohepatitis (NASH). The ultrasound-based AI program correctly detected more than 90% of NAFLD. Artificial intelligence can provide faster and more objective interpretation of NASH histology while reducing pathologist workload.

Deep learning-based visualization improves diagnosis and treatment of severe portal hypertension. ML can also predict and stratify survival odds in primary biliary disease. Furthermore, ML can and should be used to detect significant liver disease as an incidental finding in routine radiology reports.

Of particular importance is the role of artificial intelligence in predicting compensated cirrhosis at the primary care level, thereby enabling early treatment. AI performed better than doctors at using clinical information to distinguish alcoholic hepatitis from acute cholangitis.

Likewise, ultrasound-based ML algorithms outperformed experienced radiologists and produced comparable information to CT or MRI for risk prediction and stratification of early-stage HCC in patients with cirrhosis and to differentiate between benign and malignant tumors.

Machine learning for treatment planning

Liver transplantation is the final treatment for advanced liver disease. Artificial intelligence-based liver segmentation for liver transplantation helps plan liver resections and identify donor-recipient mismatches, thereby improving graft and patient survival.

While dynamic contrast-enhanced MRI is the tool of choice today, advances in hepatology will see radiogenomics used to classify tumors and predict their gene expression and mutation profiles as accurately as a pathologist with five years of experience. This reveals tumor biology and predicts treatment response.

The Deep Learning Treatment Assessment (DELTA) Liver Fibrosis Score helps monitor fibrosis treatments by predicting reductions in fibrosis severity. CT-based radiomics can predict high-risk varicose veins (HRV) in patients with compensated severe CLD, thereby limiting unnecessary surgery.

Automatic segmentation of liver tumors based on deep learning overcomes the time and skill limitations of manual segmentation. This improves tumor burden assessment and treatment planning.

Predict disease progression

A deep learning model can help prioritize patients with cirrhosis for liver transplantation by predicting one-year mortality better than traditional analysis. They can predict survival and complications after transplantation.

Artificial intelligence tools can predict the outcomes of chronic lung disease, portal hypertension, esophageal varices, and the risk of acute or chronic liver failure (ACLF).

AI automated liver and tumor segmentation accurately predicts tumor recurrence. DL can also accurately classify liver patients who are at high risk of short-term death.

The future of hepatology

Advances in hepatology based on artificial intelligence may impact the diagnosis, prognosis, and treatment of liver diseases. Artificial intelligence uses massive data sets and machine analysis to eliminate various types of bias.

Machine learning can help doctors interpret data better and faster, improve medical efficiency, and help patients improve their health. Artificial intelligence in hepatology also adds resources to practitioners in remote and low-resource settings, improving their training and promoting better patient care.

Artificial intelligence can help identify therapeutic targets using clinical, molecular and genetic data. By providing non-invasive severity assessment, ML can enhance NASH drug trial recruitment and facilitate new drug development. AI can predict drug trial results, facilitating faster and more accurate selection of drug candidates.

The introduction of artificial intelligence into clinical applications is currently limited by the urgent need for careful validation and review of the algorithms used, high-quality training and testing data sets and techniques, performance standards, and randomized clinical trials.

Because these algorithms operate automatically and at scale, “Flawed algorithms could cause harm to large numbers of patients”. The potential for data theft and privacy breaches is also high.

Once these issues are resolved, artificial intelligence can drive advances in precision and personalized medicine in the field of hepatology.

refer to

further reading