Background: The liver is the largest internal organ and the most important organ after heart and brain in the human body without which life is impossible. Diagnosis of liver disease requires a long time and sufficient expertise of the doctor. Statistical methods can be classified as an automated forecasting system and help specialists for quickly and accurately diagnose liver disease. Hidden Markov model is an intelligent and robust statistical method that has been used in present study.

Methods: The data used in this cross sectional  study collected from records of patients with  five different types of liver diseases, including cirrhosis,  liver cancer,  acute hepatitis, chronic hepatitis, and  fatty liver disease. The patients have been admitted to Afzalipour  Hospital in Kerman, Iran, from  2006  to  2013. Hidden Markov model using EM algorithm for learning was fitted to the data and for evaluating the performance of the model, criteria as accuracy, sensitivity and specificity were used.

Results: The decision, sensitivity, and specificity criteria of the model for diagnosis of each liver disease were separately calculated and the highest level criteria in diagnosis of cirrhosis of the liver were 77% decision, 82% sensitivity, and 96% specificity, and also the lowest level of diagnosis for fatty liver disease was 65% decision, 69% sensitivity and 94% specificity.

Conclusion: The results of this study indicate the potential capabilities of the Hidden Markov Model. Therefore, using Hidden Markov Model for prediction of diagnosis of liver disease is recommended