Consider a scenario where you need to fine-tune an Azure OpenAI model for a highly specialized medical task, such as diagnosing diseases from patient reports. What steps would you take to ensure the model is fine-tuned effectively and generates reliable diagnoses? Describe the fine-tuning process in detail.