The paper introduces Healthcare Copilot, a framework that enhances general large language models (LLMs) for medical consultations without requiring fine-tuning. The system consists of three main components: Dialogue, Memory, and Processing. The Dialogue component facilitates safe and effective patient interactions, including proactive questioning and ethical safety checks. The Memory component stores current and historical conversation data to improve dialogue accuracy. The Processing component summarizes dialogues and generates reports. The system was evaluated using an auto-evaluation scheme with ChatGPT, demonstrating significant improvements in inquiry capability, conversational fluency, response accuracy, and safety compared to general LLMs. Ablation studies highlight the contributions of each component. The Healthcare Copilot framework is designed to handle various medical tasks, including diagnosis, explanation, and recommendation. It is shown to be effective in real-world medical scenarios, offering high-quality consultations. The system is open-source and publicly available on GitHub. The paper also discusses related work, including other LLM-based copilots and modular prompting approaches. The results indicate that the Healthcare Copilot significantly enhances the capabilities of general LLMs for medical consultations, with GPT-4 performing the best. The framework is applicable to various general LLMs and has the potential to improve medical consultation services. However, the system is still in research phase and not a substitute for professional medical advice. The paper emphasizes the importance of ethical considerations and the need for further research to ensure the safety and effectiveness of the system.The paper introduces Healthcare Copilot, a framework that enhances general large language models (LLMs) for medical consultations without requiring fine-tuning. The system consists of three main components: Dialogue, Memory, and Processing. The Dialogue component facilitates safe and effective patient interactions, including proactive questioning and ethical safety checks. The Memory component stores current and historical conversation data to improve dialogue accuracy. The Processing component summarizes dialogues and generates reports. The system was evaluated using an auto-evaluation scheme with ChatGPT, demonstrating significant improvements in inquiry capability, conversational fluency, response accuracy, and safety compared to general LLMs. Ablation studies highlight the contributions of each component. The Healthcare Copilot framework is designed to handle various medical tasks, including diagnosis, explanation, and recommendation. It is shown to be effective in real-world medical scenarios, offering high-quality consultations. The system is open-source and publicly available on GitHub. The paper also discusses related work, including other LLM-based copilots and modular prompting approaches. The results indicate that the Healthcare Copilot significantly enhances the capabilities of general LLMs for medical consultations, with GPT-4 performing the best. The framework is applicable to various general LLMs and has the potential to improve medical consultation services. However, the system is still in research phase and not a substitute for professional medical advice. The paper emphasizes the importance of ethical considerations and the need for further research to ensure the safety and effectiveness of the system.