Abstract： Objective To evaluate the use and effectiveness of Human-Computer Interaction (HC1) -based risk prediction of diabetes among Chinese adults.Methods HCI-based risk prediction of diabetes was performed in 639 non-diabetics aged 23 to 61years old.Risk prediction results,main risk factors of diabetes and helpful suggestions were reported and used for self-management.After l-year follow-up,the participants received another assessment to find the changes of disease risk and risk factors.Non-parametric or Chi-square test was used for comparison of continuous or categorical variables,respectively.Receiver Operating Characteristic (ROC) curve was used to calculate the sensitivity and specificity of HCI.Results After1-year follow-up,the incidence of diabetes per year was1.4％,and all newly diagnosed diabetes was found in high-risk individuals.The proportion of high-risk individuals was 56.8％ and 57.9％before and after follow-up ( x2 =0.36,P ＞ 0.05 ).In comparison with baseline,average risk score of high-risk individuals was significantly declined ( 2.25 vs 2.91,Z =- 4.32,P ＜ 0.05 ).Oversized waist circumstance,higher total cholesterol (TC) and lower high-density lipoprotein cholesterol (HDL-C) was identified in 76.2％,36.2％ and 3.8％ of high risk individuals at1year,lower than those of baseline ( 87.3％,42.2％ and12.4％,respectively ; x2 values were 30.56,6.05 and 22.26,respectively; all P ＜0.05) ; although the prevalence of hypertension was higher (23.5％ vs18.1％,x2 =11.11,P＜0.05).Conclusions HCI and effective control of risk factors could prevent the development of diabetes in high risk individuals.