兔血液凝血功能诊断的改进Fisher判别法

Evaluation of coagulation function diagnosis by improved Fisher discriminantion

目的通过对经内毒素侵袭的实验对象凝血指标的检测,建立凝血指标与凝血类型之间较为准确的数学关系,从而代替血栓弹力图(TEG)检测法。方法将实验对象的凝血指标(PT、APTT、TT和FIB)通过实码加速遗传算法的投影寻踪方法赋予相应的判别权值,进而应用Fisher判别法建立指标与凝血类型间的判别关系函数,进行判别。最后通过指标筛选对判别函数进行优化。结果选用全部4项指标的判别正确率为92.00%。去除指标APTT后,判别正确率为94.8%。优化后对于低凝类型的评判准确率由最初的48.5%上升至74.29%。结论改进Fisher判别法对凝血的评判准确率高,方法简便,具有较高的应用价值。

Objective To test the coagulation indicators in the experimental subjects invaded by endotoxin to establish the mathematic relationship between the types of coagulation and the indicators of blood coagulation for replacing thromboelastogram（TEG） assay.Methods We gave the indicators of blood coagulation（including PT,APTT,TT and FIB） with the corresponding discriminant weights got by the projection pursuit（PP） method of the real coding accelerating genetic algorithm（RAGA） and established the relationship between the types of coagulation and the indicators of blood coagulation with the the Fisher discriminant.Finally,the indicators screening was adopted to optimize the discriminant function.Results The correct identification rate was 92.00% with all four indicators.Excluding the indicator APTT,the correct identification rate was 94.8%.After optimization,the correct identification rate of the low-coagulation type was increased from initial 48.6% to 74.3%.Conclusion The modified Fisher discriminant method is precise to discriminate the type of blood coagulation,which is simple and convinient with higher application value.