UNLABELLED: BACKGROUND AND RATIONALE FOR THE STUDY: Ultrasound assessment of the severity of non-alcoholic fatty liver disease (NAFLD) shows substantial observer variability. The purpose of this retrospective study is to develop a more objective, quantitative, and applicable assessment method for all physicians. MAIN RESULTS: Male gender, and increases in age, body mass index, alanine aminotransferase (ALT), triglycerides (TG), and total cholesterol (TC) were found to be significantly correlated to higher scores. The following algorithm, derived from a 3,275 member training group, for predicting the extent of fatty liver infiltration was then constructed using these parameters. In (π(1)/π(0)) = -8.360-0.065*Gender+0.010*age+0.256*BMI+0.024*ALT+0.03*TG+0.002*TC. In (π(2)/π(0))= -19.0.92+0.482*Gender+0.043*age+0.529*BMI+0.046*ALT+0.005*TG+0.005*TC. π(0): the probability of non fatty liver. π(1): the probability of degree 1 fatty liver. π(2): the probability of degree 2-3 fatty liver. π(0) + π(1) + π(2) = 1. The resulting algorithm was tested for its predictive power a 1,065 member validation group. The algorithm predicted the actual ultrasound fatty liver score in the validation group with 87.9, 14.2, and 72.6% accuracy for those with no, grade 1, and grade 2-3 fatty liver, respectively. For prediction of grade 2-3 fatty liver, its sensitivity was 70.8%, its specificity 85.2%, its positive predictive power 63.2% and its negative predictive power 88.8%. CONCLUSIONS: The algorithm developed here is fast and has substantial predictive power for grade 2-3 fatty liver. No specialized equipment or expertise is needed, and it can be easily used by the general practitioner to predict the extent of fatty infiltration in cases of NAFLD.
UNLABELLED: BACKGROUND AND RATIONALE FOR THE STUDY: Ultrasound assessment of the severity of non-alcoholic fatty liver disease (NAFLD) shows substantial observer variability. The purpose of this retrospective study is to develop a more objective, quantitative, and applicable assessment method for all physicians. MAIN RESULTS: Male gender, and increases in age, body mass index, alanine aminotransferase (ALT), triglycerides (TG), and total cholesterol (TC) were found to be significantly correlated to higher scores. The following algorithm, derived from a 3,275 member training group, for predicting the extent of fatty liver infiltration was then constructed using these parameters. In (π(1)/π(0)) = -8.360-0.065*Gender+0.010*age+0.256*BMI+0.024*ALT+0.03*TG+0.002*TC. In (π(2)/π(0))= -19.0.92+0.482*Gender+0.043*age+0.529*BMI+0.046*ALT+0.005*TG+0.005*TC. π(0): the probability of non fatty liver. π(1): the probability of degree 1 fatty liver. π(2): the probability of degree 2-3 fatty liver. π(0) + π(1) + π(2) = 1. The resulting algorithm was tested for its predictive power a 1,065 member validation group. The algorithm predicted the actual ultrasound fatty liver score in the validation group with 87.9, 14.2, and 72.6% accuracy for those with no, grade 1, and grade 2-3 fatty liver, respectively. For prediction of grade 2-3 fatty liver, its sensitivity was 70.8%, its specificity 85.2%, its positive predictive power 63.2% and its negative predictive power 88.8%. CONCLUSIONS: The algorithm developed here is fast and has substantial predictive power for grade 2-3 fatty liver. No specialized equipment or expertise is needed, and it can be easily used by the general practitioner to predict the extent of fatty infiltration in cases of NAFLD.