AI-Driven Search for Infectious Disease Research, Streamlining Information Access and Discoverability: MPOX Case Study (Preprint)

Abstract

<sec> <title>BACKGROUND</title> The exponential growth of scientific literature presents significant challenges in infectious disease outbreak management, where rapid and reliable information retrieval is crucial for evidence-based decision-making. Traditional search methods struggle to synthesize heterogeneous and rapidly evolving data sources, leading to information overload and delays in integrating critical findings into clinical and public health strategies. The AI-driven system was developed by The Global Health Network (TGHN), to ensure accessibility and applicability of resources to strengthen global health research. The monkeypox (MPOX) outbreak exemplifies these challenges, highlighting the need for AI-driven solutions to streamline knowledge discovery and access. </sec> <sec> <title>OBJECTIVE</title> Present the development and evaluation a Retrieval Augmented Generation (RAG)-based AI system for optimizing information retrieval, synthesis, and dissemination in the context of infectious disease research, using MPOX as a case study. </sec> <sec> <title>METHODS</title> A descriptive case study methodology was employed to develop a pilot AI-driven system that integrates semantic retrieval and generative capabilities. The AI-driven system is designed to enhance the accessibility and relevance of scientific literature, leveraging state-of-the-art natural language processing (NLP) and machine learning techniques to provide contextually accurate, real-time responses. The AI-driven system architecture follows a microservices-based modular design. The data pipeline incorporates sources such as PubMed, medRxiv, bioRxiv, WHO, CDC reports, and curated social media feeds. Performance evaluation included: quantitative metrics and qualitative review by infectious disease specialists. </sec> <sec> <title>RESULTS</title> The system successfully integrated traditional keyword-based retrieval with vector-based semantic search, significantly improving the relevance and contextual accuracy of retrieved information. The hybrid search model demonstrated enhanced document ranking and retrieval precision, validated through standard information retrieval metrics. The human-in-the-loop evaluation confirmed that the generated responses aligned with expert-validated sources, ensuring scientific accuracy and usability for public health professionals. The pilot implementation on MPOX research indicated the system’s scalability and adaptability to various infectious disease domains. </sec> <sec> <title>CONCLUSIONS</title> AI-driven RAG architectures offer a transformative approach to knowledge discovery and synthesis in infectious disease research. By integrating retrieval-based document indexing with generative AI, this system provides a scalable, real-time, and evidence-based solution for researchers, clinicians, and policymakers. The study underscores the need for further development to enhance multilingual support, real-time data integration, and ethical AI governance. The framework has the potential to revolutionize global outbreak preparedness, facilitating rapid, reliable, and transparent access to critical health information. </sec> <sec> <title>CLINICALTRIAL</title> Not applicable. </sec>

Ähnliche Arbeiten