基于机器学习的超声影像及SWE预测模型在早期慢性肾病中的应用OACSTPCD

Objective:To develop and validate a predictive model based on machine learning algorithms,which combines ultrasound imaging and shear wave elastography(SWE)technology,aiming to improve the diagnostic accuracy of early chronic kidney disease(CKD).Methods:This study adopts a prospective research design and includes 308 research subjects who sought medical treatment at the Affiliated Panyu Central Hospital of Guangzhou Medical University from March 2022 to November 2023.Among them,185 were chronic kidney disease patients(observation group)and 123 were control group.All research subjects underwent bilateral renal color Doppler ultrasound examination and SWE quantitative measurement,and clinical data and laboratory indicators were collected.Based on glomerular filtration rate and serum creatinine levels,patients were divided into early-stage CKD group,mid-to-late-stage CKD group,and control group.The random forest machine learning algorithm was used to analyze the parameters of different groups and establish a predictive model for early-stage CKD.Results:Single-factor analysis showed that there were statistically significant differences in parameters such as the average value of bilateral renal cortex Young modulus(YM),bilateral renal parenchymal thickness,peak systolic velocity of the right renal artery,and resistance index among different groups(P＜0.05).The random forest model identified right renal cortex YM,left renal cortex YM,and right renal artery resistance index as three key indicators for assessing the severity of CKD.The model exhibited an overall accuracy of 78.86％in CKD classification prediction,with an AUC of 0.91,sensitivity of 75.00％,and specificity of 85.54％for the early CKD group.Conclusion:The machine learning model combining color Doppler ultrasound and SWE technology performs well in assessing the severity of chronic kidney disease.The model has high accuracy in predicting early CKD,providing a non-invasive and efficient method for early diag-nosis.