Researchers at the University of California, Davis have developed a method to predict if patients diagnosed with nonalcoholic fatty liver disease are at risk for developing liver cancer using a machine learning algorithm that analyzes a variety of easily available phenotypes and risk factors.
Nonalcoholic fatty liver disease (NAFLD) is the most common chronic liver disease in the U.S., and it is typically linked to obesity, diabetes, and high blood pressure. In severe cases it can lead to liver scarring and studies show that it is a high-risk factor for hepatocellular carcinoma (HCC), the most common form of liver cancer. Determining the exact probability of high-risk patients diagnosed with NAFLD of developing cancer, without running an abundance of expensive tests, is complicated by the many risk factors contributing to HCC. Furthermore, the large quantity of data and risk factors that must be analyzed may not lead to a concise result, which could jeopardize patient outcomes. Given HCC is among the leading types of cancer mortalities, it is necessary to improve the screening process in order to detect cancer risks as soon as possible.
Researchers at the University of California Davis have developed a machine learning model trained with a large medical database that can accurately predict the chance of patients with NAFLD from developing liver cancer with high specificity and sensitivity. Some examples of risk factors that the model accounts for include results from complete blood counts, ethnicity, family history, and comorbidities such as HIV, Hepatitis B, kidney disease. The model has the potential to provide recommended treatment options that are personalized for each patient. The output of user-friendly interface also makes it easier for physicians to summarize and understand key data points along with predicted outcomes. These insights allow for selective procedure options, saving considerable time and money at a clinical scale and ultimately improving patient outcomes. In the future, this technique of training a machine learning model to analyze specific risk factors can be adapted for screening other types of cancer and life-threatening diseases.
Nonalcoholic fatty liver disease (NAFLD), hepatocellular carcinoma (HCC), liver cancer, oncology, machine learning