Hanyang University Professor Noh Yung-kyun from the Department of Computer Science and Medical Specialist Joseph Ahn from the Division of Gastroenterology in the Mayo Clinic in the U.S. co-developed the "Liver Disease Classification Technology" using machine learning. The Mayo Clinic has been selected as the top general hospital in the U.S. for 5 consecutive years and is known as the world's best hospital.

Alcoholic hepatitis and cholangitis are two different diseases but are hard to distinguish due to similar clinical results such as right upper quadrant pain, irregularities in cholestasis liver index, and inflammatory response on the entire body.

The newly developed technology is expected to help the doctors accurately distinguish and diagnose these two diseases, and provide faster treatments.

The Mayo Clinic research team has been conducting retrospective research for 10 years starting from 2010, having around 460 patients as the research subject, who are 18 years or older and have diseases of alcoholic hepatitis and cholangitis.

The research members collected 10 research laboratory variables that are included in prohaemocyte test and liver function test. The variables are △white blood cell (HBC) △hemoglobin (Hgb) △average blood cell volume (MCV) △Platelet (Plt) △AST △ALT △Alkaline phosphatase (AP) Alkaline phosphatase △Total bilirubin (Tb) △Direct bilirubin (Db) and △albumin (Alb).

Based on these variables, the results of 137 doctors and 7 supervised machine learning algorithms on distinguishing alcoholic hepatitis and cholangitis were compared.

* The 7 supervised machine learning algorithms are k-Nearest Neighborhood, Logistic regression, Supporting vector machine (SVM), Decision hierarchy tree, Naive-Bayes, Artificial neural network (ANNs), and Random forest.

As a result, the doctors showed an average of 72% accuracy in distinguishing the two diseases, while the seven supervised machine learning algorithms showed an average of 86.5%~93.6% accuracy, especially the Supporting vector machines (SVM) and the Artificial neural network (ANNs) algorithm which showed high accuracy. 

This research holds its meaning in that it allows accurate distinction between alcoholic hepatitis and cholangitis through AI technology, even in circumstances without the record of the patients or precise image medical tests.

Meanwhile, the research result was presented at the AASLD Liver Meeting in November 2020 and was appointed as the Poster of distinction. The research was also included on The Best of the Liver Meeting's Summary Slide Deck.