One Health: AI in Healthcare Summarization

The research uses a One Health text document as our main dataset. Two summarization methods were employed in this study, with the first being recursive (summary of the summary), which was repeated twice to achieve a progressively concise word count, and the second a controlled and direct summary of the original document, which was repeated to accomplish the same set word count. Performance metrics like ROUGE, BLEU, and BERT scores were calculated to assess the effectiveness of both methods. This study goes beyond automated metrics by incorporating human evaluation and assessing readability and coherence with the original document to ensure a qualitative validation of the results. Additionally, the results from all summaries generated provide a comparative analysis between traditional LLM-based summarization and conversational AI establishing performance baselines for validating clarity and coherence for effective summarization of healthcare documents.