One of the leading causes of global morbidity and mortality rates is cardiovascular diseases. While Large Language Models (LLMs) offer promising solutions in various domains, in the field of healthcare, especially where prevention, treatment planning, and patients’ improvement over time has potential in human heart diseases, as well. This chapter reviews the vital applications of both the general-purpose models like GPT-4, Bing, Gemini, as well as the domain-specific architectures like Med-PaLM, BioGPT, and more in cardiology. We will assess how the LLM processes multimodal inputs from electronic health records and wearable biosensor data in order to improve the risk prediction, automate ECG interpretation, refining of arrhythmia management, and personalize chronic disease monitoring. The empirical studies have demonstrated that LLM-guided approaches can outperform traditional risk scores in specific contexts and streamline the clinical documentation with supporting patients’ understanding via a natural language interface. With significant challenges remaining, along with ensuring data quality and representativeness, mitigating algorithmic bias, clinical validations, maintaining transparency, and seamless integration into existing workflows.