MEDIBOT: An AI-Powered Conversational Platform for Medical Knowledge Retrieval

This study introduces a more advanced energy This study introduces an intelligent, modular architecture for enhancing access to validated medical information through a conversational AI system. The MEDIBOT platform proposed herein combines Retrieval-Augmented Generation (RAG) with semantic embeddings, vector similarity search, and large language models (LLMs) to offer grounded, accurate, and context-aware medical responses to medical questions. By utilizing LangChain, FAISS, Hugging Face Transformers, and a quantized LLaMA 2 model, the system guarantees real- time responsiveness and factuality. Through this pipeline, users such as patients, students, and healthcare workers can interact with reliable, document-supported medical information through an effortless chatbot interface. MEDIBOT's power comes from its design-thinking mentality and technical stability. The design focuses on document-based grounding, controlled generation through prompt engineering, and user-friendliness through Chainlit. The framework limits misinformation and hallucinations, maximizes explainability through cited references, and enhances inclusivity through intended multilinguality. But there are a few limitations. The current implementation is English-only document corpus and does not support GPU acceleration, which can affect latency under high query rates. Also, although prompt engineering mitigates hallucination, the system does not yet offer real- time validation or peer-reviewed certainty markers in generated responses. Also, large-scale dynamic document updates still need periodic FAISS re-indexing, which can be time-consuming in production environments. Future work will address these limitations by focusing on improving scalability, multilingual capability, and content validation. Planned enhancements include: -                 Multilingual embeddings and user interface integration to enable regional language-based questions. -                 Voice-based interaction and accessibility with speech- to-text and text-to-speech feature implementation. -                 Real-time document ingestion pipeline and vector index auto-refresh usage to accommodate dynamic content refreshes. -                 Research into federated learning methods for decentralized chatbot training across several clinical or institution-based contexts. -                 Confidence scoring and citation-based validation integration to enhance trust in AI response even further. -                 Cloud-native deployment with GPU support to improve concurrency and reduce inference latency. In conclusion, MEDIBOT illustrates that retrieval-enhanced conversational AI, when paired with user-centric design and domain-specific document processing, can revolutionize digital health accessibility. As subsequent versions evolve in performance, adaptability, and inclusivity, MEDIBOT is poised to become a trusted companion in medical education, patient empowerment, and clinical assistance—setting the groundwork for more intelligent, safer, and more human- focused AI in healthcare.

[1]       Forbes, "Google Trusts DeepMind AI to Manage Data Centre Cooling," Aug. 18, 2018. [Online]. Available: https://www.forbes.com [2]       Quantum Zeitgeist, "Deepmind AI Cuts Google Data Center Cooling Bill By 40%," Feb. 27, 2025. [3]      The Guardian, "Google uses AI to cut data centre energy use by 15%," Jul. 20, 2016. [4]      JATIT, "Deep Learning-Driven Forecasting Models for IoT," Mar. 31, 2025. [5]      ScienceDirect, "Deep CNN and LSTM Approaches for Efficient Workload Prediction," 2024. [6]      IEEE Xplore, "AI-Powered Healthcare Monitoring using Edge Devices and IoT," Oct. 2023. [7]         Springer, "Smart Medical Systems: An IoT-Based Approach to Real-Time Patient Monitoring," Feb. 2022. [8]     Elsevier, "Energy-Efficient Cloud Resource Management in AI Healthcare Applications," Jul. 2024. [9]          MDPI Sensors, "A Survey on Wearable Health Monitoring Systems and Their Integration with AI," Dec. 2023. [10]       Nature Medicine, "Artificial Intelligence in Clinical Decision Support Systems: Present and Future," Jan. 2022.