Efficient pumping of whole blood is an essential task in biomedical engineering, especially for point-of-care diagnostics using lab-on-a-chip devices. Alternating current (AC) electrokinetics have been widely used for several different applications among which pumping fluids using the precisely controlled electric field without any moving mechanical parts is significant. Due to its high conductive characteristic, it is difficult to drive the blood flow using the AC electroosmosis phenomenon because the electric double layer is highly compressed. Fortunately, the AC electrothermal (ACET) phenomenon occurs due to the variation of temperature-dependent permittivity and conductivity caused by Joule heating effects or other heat sources making it powerful for driving high electrical conductivity physiological fluids in biomedical devices. Compared with Newtonian fluids like saline solutions or urine, the non-Newtonian rheological nature and AC frequency-dependent dielectric property of blood make its ACET driving mechanism more complicated and attractive. In this paper, ACET induced blood flow in the 3D microfluidic channel is modeled by the lattice Boltzmann method accelerated using graphics processor units. The Carreau-Yasuda model is applied to simulate the shear-thinning behavior of blood flow, and its electrothermal pumping efficiency is investigated with respect to the AC electrode configuration, AC voltage magnitude, and AC signal frequency by comparing it with the ACET pumping of Newtonian fluids using scaling law analysis. The results demonstrate that the ACET phenomenon is effective for pumping non-Newtonian whole blood flow in microfluidic devices with ring electrodes which will contribute to the point-of-care diagnostic of bacterial bloodstream infections or rapid detection of circulating tumor cells.