Large language models (LLM) quickly transitioned from an academic niche to a mainstream technology, making them an integral part of the growing field of generative artificial intelligence. Generative models still pose a significant challenge due to non-determinism causing varying output with small prompt adjustments. This paper introduces Concept Type Prompt Patterns: a formalized prompting strategy designed to enhance the quality of LLM outputs by directing the model toward more correct and concise responses. The Concept Type Prompt Patterns comprise structured prompting templates that aim to guide the LLM in generating specific types of textual outputs. The Concept Type Prompt Patterns were applied to automated medical reporting in a case study in the field of healthcare. By validating the generated outputs with a mixed-method evaluation approach using quality metrics and human evaluation, this research demonstrates that the Concept Type Prompt Patterns offer a structured approach that improves LLM performance, addressing key challenges in prompt engineering and enhancing the reliability of LLM-based systems in healthcare.