The utility of an artificial intelligence model based on decision tree and evolution algorithm to evaluate steatotic liver disease in a primary care setting.

Many ways of classifying steatotic liver disease (SLD) with metabolic conditions have been proposed. Thus, SLD-related variables were verified using a decision tree. We tested if the suggested components of the actual classification (metabolic dysfunction-associated steatotic liver disease, MASLD) are also present in young and middle-aged adults. In a cross-sectional study involving 6,839 adults (median age: 46 years, 69.5% men) in a primary care setting, a decision tree was created to determine potential clinical and laboratory variables related to SLD. The odds ratio (OR) with a respective 95% confidence interval (95%CI) was calculated for both sexes. SLD frequency was 26.6% (23% in men). More variables and with higher ORs for the association with SLD were identified in women: category 1 (body mass index (BMI) ≥29 kg/m2, age <51 years, high-sensitivity C-reactive protein (hs-CRP) ≥0.195 mg/dL): OR=10.9, 95%CI: 4.40-26.6; category 2 (BMI <9 kg/m2, metabolic syndrome (MS), age ≥50 years, neck circumference (NC) ≥36 cm): OR=8.1, 95%CI: 2.2-29.9; and category 3 (BMI ≥29 kg/m2, age <51 y-old, dyslipidemia, high-density lipoprotein cholesterol (HDL-c) <42 mg/dL): OR=4.7, 95%CI: 2.20-10.7. For men: category 1 (waist circumference (WC) ≥101 cm, alanine aminotransferase (ALT) <28 mg/dL, glycated hemoglobin (HbA1c) ≥5.7%): OR=4.7, 95%CI: 2.8-7.9; and category 2 (WC ≥101 cm, ALT ≥28 mg/dL): OR=3.2, 95%CI: 2.5-4.0). The decision tree identified more variables related to SLD, particularly in women, such as age of more than 50 years, elevated hs-CRP, and NC≥36 cm than variables related to MASLD.