Abstract In order to have research on the deformation characteristics and mechanical properties of human red blood cells (RBCs), finite element models of RBC optical tweezers stretching and atomic force microscope (AFM) indentation were established. Non-linear elasticity of cell membrane was determined by using the neo-Hookean hyperelastic material model, and the deformation of RBC during stretching and indentation had been researched in ABAQUS, respectively. Considering the application of machine learning (ML) in material parameters identification, ML algorithm was combined with finite element (FE) method to identify the constitutive parameters. The material parameters were estimated according to the deformation characteristics of RBC obtained from the change of cell diameter with stretching force when RBC was stretched. The non-linear relationship between material parameter and RBC deformation was established by building a FE-model. The FE simulation of RBC stretching was used to construct the training set and the neural network trained by a large number of samples was used to predict the material parameter. With the predicted parameter, FE simulation of RBC under AFM indentation to explore the local deformation mechanism was completed.

中文翻译：

---

利用神经网络方法和有限元分析对变形红细胞进行数值模拟

摘要 为了研究人红细胞（RBC）的变形特性和力学性能，建立了红细胞光镊拉伸和原子力显微镜（AFM）压痕的有限元模型。利用neo-Hookean超弹性材料模型确定细胞膜的非线性弹性，在ABAQUS中分别研究了RBC在拉伸和压痕过程中的变形。考虑到机器学习（ML）在材料参数识别中的应用，ML算法结合有限元（FE）方法识别本构参数。根据RBC拉伸时细胞直径随拉伸力的变化得到RBC的变形特性，估计材料参数。通过建立有限元模型，建立了材料参数与 RBC 变形之间的非线性关系。利用RBC拉伸的有限元模拟构建训练集，利用大量样本训练的神经网络预测材料参数。使用预测的参数，完成了 AFM 压痕下 RBC 的有限元模拟，以探索局部变形机制。