This research systematically evaluates the accuracy and hallucination rates of three key LLM enhancement techniques across critical domains.

• RAG significantly reduces hallucinations by incorporating external knowledge
• LoRA provides efficient fine-tuning with modest accuracy improvements
• DoRA demonstrates superior performance by optimizing weight decomposition
• Combined approaches show promising results for specialized applications

For medical applications, this research offers critical guidance on which techniques minimize potentially harmful misinformation, essential for clinical decision support, patient education, and healthcare documentation systems.

Hallucinations and Truth: A Comprehensive Accuracy Evaluation of RAG, LoRA and DoRA