This study explores the use of large language models (LLMs) to automate the generation of discharge notes for cardiac patients, aiming to improve clinical efficiency and documentation accuracy. The research evaluates the effectiveness of various LLMs, with Mistral-7B showing the most promising results. The model was trained on a comprehensive dataset from a cardiology department, including detailed patient records and physician assessments. The discharge notes generated by Mistral-7B were evaluated both quantitatively and qualitatively, with high scores in clinical relevance, completeness, readability, and utility. Quantitative metrics such as ROUGE, BLEU, BERT Score, and Perplexity confirmed the model's ability to produce accurate and coherent discharge notes. Qualitative assessments by medical experts further validated the model's performance, highlighting its accuracy in capturing patient information and its usefulness in clinical decision-making and follow-up care planning. The study demonstrates the potential of specialized LLMs in streamlining healthcare documentation and improving patient care. However, challenges remain, including the need for domain-specific evaluation metrics and the variability in progress notes formatting. Future research should focus on expanding the application of these models to other medical specialties and incorporating multi-modal data sources to enhance their capabilities. The integration of advanced techniques such as attention mechanisms and multimodal fusion could further improve the models' performance. Overall, this study highlights the significant potential of LLMs in transforming healthcare documentation practices and supporting better patient outcomes.This study explores the use of large language models (LLMs) to automate the generation of discharge notes for cardiac patients, aiming to improve clinical efficiency and documentation accuracy. The research evaluates the effectiveness of various LLMs, with Mistral-7B showing the most promising results. The model was trained on a comprehensive dataset from a cardiology department, including detailed patient records and physician assessments. The discharge notes generated by Mistral-7B were evaluated both quantitatively and qualitatively, with high scores in clinical relevance, completeness, readability, and utility. Quantitative metrics such as ROUGE, BLEU, BERT Score, and Perplexity confirmed the model's ability to produce accurate and coherent discharge notes. Qualitative assessments by medical experts further validated the model's performance, highlighting its accuracy in capturing patient information and its usefulness in clinical decision-making and follow-up care planning. The study demonstrates the potential of specialized LLMs in streamlining healthcare documentation and improving patient care. However, challenges remain, including the need for domain-specific evaluation metrics and the variability in progress notes formatting. Future research should focus on expanding the application of these models to other medical specialties and incorporating multi-modal data sources to enhance their capabilities. The integration of advanced techniques such as attention mechanisms and multimodal fusion could further improve the models' performance. Overall, this study highlights the significant potential of LLMs in transforming healthcare documentation practices and supporting better patient outcomes.