This study explores the use of large language models (LLMs) to automate the generation of discharge notes for cardiac patients, aiming to enhance clinical efficiency and improve patient care. The research leverages a substantial dataset from the Cardiology Department of Asan Medical Center, focusing on the efficiency and accuracy of LLMs in creating comprehensive discharge notes. Among the models evaluated, Mistral-7B stands out for its ability to generate discharge notes that are clinically relevant, complete, readable, and useful for informed decision-making and care planning. The study employs both quantitative metrics such as ROUGE, BLEU, BERT Score, and Perplexity, and qualitative evaluations by medical experts to assess the performance of the models. The results demonstrate the potential of LLMs in streamlining healthcare documentation and improving the continuity of care. However, the study also highlights limitations, such as the lack of domain-specific evaluation metrics and the need for standardized input data. Future research directions include developing more specialized evaluation metrics, expanding the application to other medical specialties, and integrating multi-modal approaches to enhance the models' performance. Overall, the study provides a significant step towards leveraging AI to revolutionize healthcare documentation practices.This study explores the use of large language models (LLMs) to automate the generation of discharge notes for cardiac patients, aiming to enhance clinical efficiency and improve patient care. The research leverages a substantial dataset from the Cardiology Department of Asan Medical Center, focusing on the efficiency and accuracy of LLMs in creating comprehensive discharge notes. Among the models evaluated, Mistral-7B stands out for its ability to generate discharge notes that are clinically relevant, complete, readable, and useful for informed decision-making and care planning. The study employs both quantitative metrics such as ROUGE, BLEU, BERT Score, and Perplexity, and qualitative evaluations by medical experts to assess the performance of the models. The results demonstrate the potential of LLMs in streamlining healthcare documentation and improving the continuity of care. However, the study also highlights limitations, such as the lack of domain-specific evaluation metrics and the need for standardized input data. Future research directions include developing more specialized evaluation metrics, expanding the application to other medical specialties, and integrating multi-modal approaches to enhance the models' performance. Overall, the study provides a significant step towards leveraging AI to revolutionize healthcare documentation practices.