兰州医学院学报
蘭州醫學院學報
란주의학원학보
JOURNAL OF LANZHOU MEDICAL COLLEGE
2004年
1期
11-16
,共6页
糖尿病决策支持%决策模型%决策算法
糖尿病決策支持%決策模型%決策算法
당뇨병결책지지%결책모형%결책산법
decision support model%diabetes decision support%decision algorithm
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±0.05,n=14)和控制目标(r=0.95±0.06,n=14).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0.21.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.
目的研究多因素箇性化糖尿病決策支持繫統,改善醫療效果,解決治療中的箇性化難題.方法首次利用專傢混閤算法整閤瞭普通藥物、胰島素、膳食、體能等多參數和多繫數,為患者和專傢提供閤適的決策指導.描述瞭專傢混閤算法在糖尿病治療中的應用實施步驟.建立瞭通用臨床決策支持、糖尿病決策支持箇性化糖尿病決策支持模型,用箇性化模型對每天的治療提供指導.結果該繫統的決策結果很接近訓練數據(r=0.97±0.05,n=14)和控製目標(r=0.95±0.06,n=14).經驗法指導治療所得結果的標準差為1.81,箇性化決策所得決策結果的標準差為0.21.經驗法與決策結果之間的標準差為0.783.結論決策所得結果顯示具有更好的均衡性,證明專傢混閤算法建立的箇性化糖尿病決策支持繫統是彊有力的糖尿病決策工具,它可處理多繫數和多參數決策數據.
목적연구다인소개성화당뇨병결책지지계통,개선의료효과,해결치료중적개성화난제.방법수차이용전가혼합산법정합료보통약물、이도소、선식、체능등다삼수화다계수,위환자화전가제공합괄적결책지도.묘술료전가혼합산법재당뇨병치료중적응용실시보취.건립료통용림상결책지지、당뇨병결책지지개성화당뇨병결책지지모형,용개성화모형대매천적치료제공지도.결과해계통적결책결과흔접근훈련수거(r=0.97±0.05,n=14)화공제목표(r=0.95±0.06,n=14).경험법지도치료소득결과적표준차위1.81,개성화결책소득결책결과적표준차위0.21.경험법여결책결과지간적표준차위0.783.결론결책소득결과현시구유경호적균형성,증명전가혼합산법건립적개성화당뇨병결책지지계통시강유력적당뇨병결책공구,타가처리다계수화다삼수결책수거.
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.