人工神经网络在Horn’s法LD_(50)试验中的应用

Application of artificial neural networks in LD_(50) test using Horn's method

目的 :探索LD50 计算方法新思路。方法 :以 4组动物的死亡数和给药剂量作为输入量 ,用神经网络计算LD50 ,并对神经网络的隐层节点数进行选择 ,以提高计算精度。结果 :对急性毒作用带较宽的化学物质 ,应用人工神经网络方法计算其LD50 ,能客观地反映实际情况 ,且节省人力和动物。结论 :人工神经网络方法计算结果可以对Horn’s法LD50

Aim:To study a new idea of computing LD 50 for practical application.Methods: LD 50 was computed by artificial neural networks with the number of died animals in 4 groups and the given doses as inputs. In order to improve the accuracy, the number of the hidden units of the neural networks was selected.Results: Computing LD 50 by artificial neural networks was practical for the chemical material with broad acute toxic zone.Conclusion: The results of computing with neural networks can be a supplement for the Horn'method.