基于遗传算法的尖牙冠宽的预测

Prediction of mesiodistal crown diameters of canine based on genetic algorithms method

目的:建立尖牙与其他牙冠宽的相关方程,探索一种准确预测未萌尖牙宽度的方法。方法:用游标卡尺测量牙列模型118付(男性50付,女性68付),利用遗传算法(GAS法)建立尖牙与其他牙冠宽的关系方程:优化方程系数时采用的样本称为模拟样本(112付),其他作为方程检验的样本称为检验样本(6付),并比较其与逐步回归分析法的精度。结果:建立尖牙与其他牙冠宽的多元相关方程:MU=0.0511U1+0.2164U2+0.2780U4+0.0407U5+0.2853U6+0.8321①;ML=0.2998U1+0.1294U2+0.3912U4+0.0088U5+0.0791U6+0.9839②;FU=0.1419U1+0.1741U2+0.3258U4+0.0412U5+0.093U6+1.7355③;FL=0.2796U1+0.3750U2+0.2968U4+0.0268U5+0.0043U6+0.6030④。GAS法与逐步回归分析法比较,模拟精度的平均误差最小(P<0.05);实际预测的结果进行比较,男性的绝对误差平均值GAS法小于逐步回归分析法。结论:利用此方程预测未萌尖牙的冠宽,其精度优于逐步回归分析法,对于治疗方案的制定有临床参考价值。

Objective To establish an equation between canine and other permanent teeth, and probe a new method to predict the sum of the mesiodistal diameters of unerupted permanent canines. Methocls Samples consisted of 118 dental casts Were obtained from Chinese patients （50 males, 68 females, respectively）. Mesiodistal tooth diameters were measured by a vernier caliper. GAS was used to establish the equation between canine and other permanent teeth, and compared with stepwise regression analysis. Those were modelling samples that optimized equations coefficient （112 samples）. Others were inspection samples （6 samples）. Results The equation between canine and other permanent teeth was established.MU=0.0511U1＋0.2164U2＋0.2780U4＋0.0407U5＋0.2853U6＋0.8321①;ML=0.2998U1＋0.1294U2＋0.3912U4＋0.0088U5＋0.0791U6＋0.9839②;FU=0.1419U1＋0.1741U2＋0.3258U4＋0.0412U5＋0.093U6＋1.7355③;FL=0.2796U1＋0.3750U2＋0.2968U4＋0.0268U5＋0.0043U6＋0.6030④ . Compared with stepwise regression analysis, the mean error of precision by GAS was less （P〈0.05）. Comparing the practical predictional diameters between two methods, the absolute mean error of male was less by GAS. Conclusion The equation by GAS, which could predict diameters of unerupted canines, shows more accuracy than stepwise regression analysis. It is easy to use and valuable for orthodontic design proposal.