Applying Radiomics and Deep Learning to Investigations into Seafarers' Health Status

Abstract

This retrospective study on 7232 male seafarers(June 2022 -- Nov 2023), explored the occurrence of fatty liver disease and gallstones and pointed out its related risk elements, moreover examined the worth of using DL-based radiomics in intelligent examination and medical diagnosis of the said hepatobiliary diseases. Clinical data (BMI, blood lipids) and abdominal ultrasound images were analyzed via traditional statistical methods (chi-squared test, correlation analysis, t-test). A DL framework integrating U-Net segmentation and MobileNetV2 classification was developed to automate region-of-interest (ROI) extraction, extract high-dimensional radiomic features, and fuse clinical/radiomic data for dual-disease prediction. Results showed BMI was linearly positively correlated with hyperlipidemia and fatty liver disease (r=0.98, P<0.05). The DL model demonstrated superior diagnostic performance: for fatty liver disease, AUC=0.93, accuracy=90.2%, recall=88.5%, and specificity=89.8% (significantly higher than manual ultrasound, AUC=0.79, P<0.05); for gallstones, AUC=0.89, accuracy=87.6%, and recall=85.3%. Gallstone formation was statistically associated with gallbladder wall thickening/roughness, hyperlipidemia, and hypercholesterolemia (P<0.005).Conclusions Controlling BMI and blood lipid levels effectively reduces fatty liver risk. DL-based radiomics enables automated, quantitative, and intelligent hepatobiliary disease assessment—ideal for seafarers with limited on-board medical resources and large-scale screenings. Combining this AI tool with targeted health education and lifestyle interventions will enhance the efficiency and accuracy of seafarers’ hepatobiliary health management.