Hong Kong Med J 2025;31(Suppl 7):S4-8
Machine learning models for estimating risks of hepatic complications in patients with nonalcoholic fatty liver disease: risk stratification and treatment recommendation (abridged secondary publication)
TCF Yip1, GLH Wong1, VWS Wong1, HLY Chan2, YK Tse1, PC Yuen3, QX Tan4
1 Department of Medicine and Therapeutics, The Chinese University of Hong Kong, Hong Kong SAR, China
2 Department of Internal Medicine, Union Hospital, Hong Kong SAR, China
3 Department of Computer Science, Hong Kong Baptist University, Hong Kong SAR, China
4 Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong SAR, China
 
 
  1. A deep learning model incorporating domain adaptation techniques was developed using clinical, laboratory, and medication data to predict liver-related complications in patients with nonalcoholic fatty liver disease (NAFLD).
  2. The deep learning model demonstrated accuracy in both a development cohort of patients with NAFLD and an independent validation cohort of patients with type 2 diabetes and probable NAFLD.
  3. The deep learning model can guide referrals, further assessments, and intervention recommendations in patients with type 2 diabetes at risk of NAFLD.