一种新型糖尿病远程医疗个性化决策支持系统(英文)

A novel individual decision support system in diabetic telecare

目的 研究多因素个性化糖尿病决策支持系统 ,改善医疗效果 ,解决治疗中的个性化难题。方法 首次利用专家混合算法整合了普通药物、胰岛素、膳食、体能等多参数和多系数 ,为患者和专家提供合适的决策指导。描述了专家混合算法在糖尿病治疗中的应用实施步骤。建立了通用临床决策支持、糖尿病决策支持个性化糖尿病决策支持模型 ,用个性化模型对每天的治疗提供指导。结果 该系统的决策结果很接近训练数据 (r =0 .97± 0 .0 5 ,n =1 4 )和控制目标 (r=0 .95± 0 .0 6,n =1 4 ) .经验法指导治疗所得结果的标准差为 1 .81 ,个性化决策所得决策结果的标准差为 0 .2 1 .经验法与决策结果之间的标准差为 0 .783。结论 决策所得结果显示具有更好的均衡性 ,证明专家混合算法建立的个性化糖尿病决策支持系统是强有力的糖尿病决策工具 ,它可处理多系数和多参数决策数据。

Objective The aim of this study is to set up a multi-factors model of individual diabetic decision support system to improve the effects of diabetic care and solve individual problems in therapy. Methods The Mixture of Experts algorithm was first time to integrate multiple coefficients and parameters about general drugs, insulin, and diet, exercises, to give a suitable advice result to be considered by patients and experts. The application of the Mixture of Experts algorithm in decision support of diabetic therapy and its implementation were described. The generic clinical decision support model and the decision support model for diabetes were built. The model of individual diabetic decision support for diabetic therapy and care was set up to give daily therapeutic advice. Results The results of advice were very close to training data ( r =0.97±0.05, n =14) and the control aim ( r =0.95±0.06, n =14). The data from the experience had a standard deviation (SD) about 1.81 and the results from the advice showed that SD was 0.21. The SD between results from experience and from advice was 0.783. Conclusion The results from advice are more balanced and show that the individual diabetic decision support model with Mixture of Experts is robust methods, which can deal with multiple coefficients and parameters in application of individual diabetic decision support system.