Traditional air dehumidification relies on condensation through a vapor compression refrigeration system to remove the excess vapor moisture from the air. However, this process necessitates excess cooling, leading to energy inefficiency. In contrast, researchers have explored an alternative approach involving liquid desiccant-based dehumidification, which removes air moisture through an absorption process without excessive cooling. Recent investigations have delved into a novel liquid desiccant-based dehumidification LDDH configuration coupled with a Heat Pump using internally cooled dehumidifiers. Internally cooled dehumidifiers are a heat and mass transfer device involving three fluids: humid air, liquid desiccant, and refrigerant fluid. The intricate interplay between heat and mass transfer in the internally cooled dehumidifiers requires discretization methods for solving the complex governing equations. These models are computationally intensive and demand a comprehensive characterization of the device. Recognizing these limitations, there is a need for more suitable models that can be applied in system-level simulation for the new heat pump-coupled internally cooled dehumidifier system with control systems. The study aims to bridge the gap by employing a machine-learning approach to model the internally cooled dehumidifier. Artificial Neural Network-based models for the internally cooled dehumidifier and regenerator were successfully trained and validated using the data generated by an experimentally validated finite differences model. The artificial neural networks-based models were subsequently integrated into Modelica and incorporated into a comprehensive energy simulation that includes the heat pump and internally cooled dehumidifier. The simulation results show that the system can successfully reach the desired supply air temperature and humidity conditions and reach a favorable average system COP for the cooling season of 5.9, and maximum system COP values of 7.7.