What an inspiring week at the World Vaccine Congress 2025 in Amsterdam. The energy was palpable, driven by a unified mission: how do we accelerate vaccine R&D and deployment while maintaining the highest standards of evidence and safety?
I was honored to co-present our poster, “AI-Powered Data Extraction for Streamlining Systematic Literature Review Delivery: A COVID-19 SLR Case Study,” alongside Jack Said from Pfizer.
Our work highlighted how DistillerSR’s Smart Evidence Extraction (SEE), verified by a Human-in-the-Loop (HITL) approach, significantly reduced the time required for data extraction in a complex COVID-19 SLR while maintaining impeccable data accuracy.
The Challenge: Overcoming the Extraction Bottleneck
The data extraction phase of a systematic literature review (SLR) is historically a manual bottleneck—labor-intensive, time-consuming, and susceptible to human error. Our poster described the application of purpose-built Generative AI, integrated into a rigorous HITL workflow, to enhance the efficiency of data extraction.
Key Findings: A Shift in Operational Efficiency
The results demonstrate a measurable impact on research timelines:
- Substantial Time Savings: Total extraction time was reduced from 85 minutes per study (manual) to just 55 minutes using SEE.
- Total Efficiency Gain: Over the course of the study, this AI-driven approach saved approximately 9.5 hours of high-level researcher time.
- Precision and Accuracy: SEE accurately captured complex data, including study designs, population demographics, and specific COVID-19 variant information.
Trust Through “Expert-in-the-Loop” Oversight
For regulatory-grade research, “black box” AI is not an option. Our approach ensures 100% traceability and audit-readiness. While the AI accelerates the identification and extraction of data points, human oversight remains a fundamental tenet; researchers remain the final authority, validating or refining the AI’s suggestions to ensure regulatory compliance.
While SEE occasionally encountered difficulties with highly complex results tables—requiring manual capture—the overall reduction in administrative burden was significant.
The DistillerSR Advantage
By moving away from spreadsheets and manual data extraction, organizations gain significant efficiencies through AI-enabled workflows. As this case study confirms, purpose-built AI does not replace the researcher; it empowers them. By automating the administrative burden, we allow experts to focus on what matters most: high-value analysis and faster delivery of research.
