人工神经网络对心肌细胞动作电位时程恢复特性的拟合

目的 研究当前动作电位时程（APD）的和先前心动周期中舒张间隔（DI）、APD之间复杂的恢复特性关系。方法用不同协议起搏Luo-Rudy模型，描绘当前APD和其前一心动周期中DI之间的直角坐标曲线，用人工神经网络（ANN）拟合这些曲线。结果不同起搏协议描绘的曲线差别很大，当前APD和其前一个心动周期中的DI之间不存在一一对应的函数关系；ANN可以很高精度地拟合这些不同起搏协议描绘的曲线。结论 APD和其前面3个心动周期中的APD、DI之间存在非常复杂的恢复特性关系，ANN可足够精确地表达这种关系。

Fitting APD Restitution with Artificial Neural Network

Objective To study the complex restitution relation between the present action potential duration (APD) and its previous diastole interval (DI) and APD. Method The Luo-Rudy model was paced with different protocols to draw the coordinate curves between the present APD and its first previous DI. These curves were fit with the artificial neural network (ANN). Result Different pacing protocols caused different coordinate curves, and there was no one-to-one relation between pre-sent APD and its first previous DI. ANN fits these curves satisfactorily. Conclusion There is a complex relation between the present APD and the APDs and DIs of the first three previous beats. This relation can be modeled accurately with ANN.