Gallium nitride (GaN) high electron mobility transistors (HEMTs) have gained significant attention for their potential use in high-power amplifiers and high-frequency switching applications. However, due to their high power density, self-heating in GaN HEMTs can cause device performance degradation and shorten device lifespan. To develop effective thermal management strategies, it is crucial to gain a clear understanding of the heat generation and transport processes in the device, to accurately predict the junction temperature. In this study, we have employed a combined approach of technology computer-aided design (TCAD) and phonon Monte Carlo (MC) simulations to investigate self-heating in GaN HEMTs. TCAD simulations are utilized to generate heat source distributions in the channel, which serve as phonon emission sources for the MC simulations. Our simulation results reveal that bias-dependent heat generation in the channel has a significant impact on the thermal spreading process, as well as the phonon ballistic effects. To predict the device junction temperature with respect to bias-dependent phonon transport in the HEMT, we have proposed a two-thermal-conductivity model based on a two-heat-source model. Our proposed model can be easily coupled with finite element method (FEM)-based thermal analysis and applied to devices with various geometries, without requiring time-consuming multiscale electrothermal simulations.