A descriptive study of fatty liver disease detection using machine learning

Hepatic steatosis, which is widely familiar as fatty liver disease (FLD), is a condition in which too much fat is formed inthe liver. Though fatigue-inducing and occasionally painful, it is easily treatable through medicine and lifestyle changes. However, hepatic steatosis is linked with some life-threatening complications such as cirrhosis, liver cancer, fibrosis and esophageal varices. Since FLD is an early indicator of these complications, the odds of patient mortality may increase substantially if it is left untreatedor undetected during diagnosis. Machine learning is being applied in various classification-related problems to reduce human error, which can prove to be disastrous in healthcare. This paper attempts to survey various recently published literature that haveproposed different approaches to detecting FLD using machine learning techniques. Some of the surveyed techniques utilize digitalimage processing methods to analyze the medical images for identifying the desired features out of them, while other techniques use medical records to identify the risk predictors. Any of the machine learning models proposed in the literature will use those quantified features to predict FLD and possibly assist medical professionals, both experienced and inexperienced, in determining the high-risk patients. The implementation of these techniques in a clinical setting would greatly benefit the patients in receiving an accurate diagnosis swiftly. Along with this, the health care institutions will also gain from this in terms of reduced costs and solving the lack of manpower.