Breaking the Evidence Bottleneck: A Hybrid Artificial Intelligence and Human Pipeline to Map the over Thirty-Five Years Immunomodulatory Legacy of Lactobacillus rhamnosus GG

Background Lactobacillus rhamnosus GG (LGG) is one of the most studied probiotic strains, with an immunological fingerprint that has evolved over three decades of research. The exponential growth of biomedical literature has created a scalability crisis exceeding human synthesis capacity. Aim and Study Design This study aimed to elucidate LGG's immunomodulatory role using a zero-cost, high-speed, hybrid human-AI workflow, leveraging free Large Language Models as a scalable screening tool accessible to any researcher regardless of institutional affiliation. Methods An operative pipeline based on four specialized prompts was developed to interrogate three LLM architectures (Gemini 3 Flash, Perplexity Academic, and Claude 4.6 Sonnet), screening 35 years of PubMed literature (1989–2026). Performance was benchmarked against a human-validated Gold Standard using Sensitivity, Specificity, and Spearman Correlation, with a "Semantic Floor" analysis investigating False Negatives. Results A paradigm shift emerged: historical studies (pre-2019) focused on mucosal barrier integrity and IgA production, while recent research (2019–2026) emphasizes complex molecular signaling (IL-22, Type-I Interferons, Inflammasome modulation). A consistent core fingerprint persisted, characterized by TNF-α/IL-6 suppression and IL-10/IFNg upregulation. Gemini 3 Flash showed superior sensitivity (0.93–0.97), while Perplexity Academic achieved near-perfect specificity (0.99) due to its Retrieval-Augmented Generation (RAG) architecture which mitigated hallucinations. The Spearman Heatmap revealed a correlation coefficient near zero between Gemini and Perplexity sessions, statistically validating their architectural complementarity. Conclusions LGG research has shifted from mucosal barrier protection to complex molecular signaling. Synergy between different LLM architectures allows researchers to act as AI-orchestrators, focusing on critical interpretation rather than mechanical extraction.