【摘要】：目的通過數(shù)學方法建立輕度認知功能障礙患者24個月預(yù)后的判別評估模型。方法通過對152例輕度認知功能障礙患者的27項臨床指標的分析,分別運用貝葉斯判別分析、決策樹、BP神經(jīng)網(wǎng)絡(luò)、模糊C均值聚類及K均值聚類進行評估預(yù)測,選擇準確率較高的判別模型。結(jié)果通過差異性檢驗,根據(jù)P0.01共篩選出6項指標,建立5種評估預(yù)測模型,經(jīng)過100次隨機抽樣診斷模擬,得到貝葉斯判別分析在輕度認知功能障礙患者預(yù)后評估的平均正確率達74.86%,最高正確率達到84.21%。結(jié)論本研究建立的數(shù)學模型能提高輕度認知功能障礙患者24個月預(yù)后的評估準確率。
[Abstract]:Objective to establish a 24-month prognostic model for patients with mild cognitive impairment by mathematical method. Methods based on the analysis of 27 clinical indexes in 152 patients with mild cognitive impairment, Bayesian discriminant analysis, decision tree BP neural network, fuzzy C-means clustering and K-means clustering were used to evaluate and predict. Select the discriminant model with high accuracy. Results through the difference test, six indexes were screened out according to P0.01, 5 kinds of evaluation and prediction models were established, and 100 random sampling diagnoses were simulated. The average correct rate of Bayesian discriminant analysis in prognosis evaluation of patients with mild cognitive impairment was 74.86 and the highest accuracy was 84.21. Conclusion the mathematical model established in this study can improve the accuracy of 24 months prognosis in patients with mild cognitive impairment.
【作者單位】： 第三軍醫(yī)大學學員旅5營;第三軍醫(yī)大學生物醫(yī)學工程系生物醫(yī)學材料學教研室;第三軍醫(yī)大學生物醫(yī)學工程系數(shù)學與生物數(shù)學教研室;
【基金】：重慶市自然科學基金(CSTC2013jcyjA10041)~~
【分類號】：R749.1
，

本文編號：2190998