Accelerating Systematic Literature Reviews (SLRs) for Faster GVD Development
The Challenge
A leading global biopharmaceutical company, recognized for its robust evidence-generation practices, wanted to streamline the creation of Systematic Literature Reviews (SLRs), a critical foundation for its Global Value Dossier (GVD). While their internal experts ensured scientific rigor, the manual SLR process was consuming significant time and resources, often delaying downstream deliverables such as value stories and payer submissions. The client’s evidence and medical affairs teams were facing:
- Long SLR timelines: Each review cycle could take 10–12 weeks due to manual data extraction and summarization.
- Inconsistent data representation: Different therapeutic areas followed varied extraction formats, making cross-indication analysis difficult.
- Limited bandwidth of experts: Highly skilled reviewers were spending valuable time on repetitive screening and data-cleaning tasks.
They wanted to explore AI-assisted automation to accelerate literature reviews without compromising scientific accuracy or regulatory traceability
Our Approach
Agilisium partnered with the client’s Global Evidence function to deploy its Content Generation platform, purpose-built for scientific and regulatory documentation.
The engagement was designed to augment not replace the client’s existing expertise.
Key focus areas
- SLR Automation Framework: Using Content Generation’s LLM-driven pipeline to automate search string creation, abstract screening, and data extraction aligned to PICO standards
- Human-in-the-Loop Validation: Ensuring that AI-generated outputs were reviewed and approved by subject matter experts for scientific credibility
- Structured Evidence Outputs: Delivering standardized evidence tables and narrative summaries directly compatible with GVD templates
Our Solution
Automated Search Strategy Generation:
Content Generation drafted database-specific search strategies across PubMed, Embase, and Cochrane, which were then reviewed by client librarians, reducing initial setup time by over 60%
AI-Assisted Abstract Screening:
The tool ranked and prioritized abstracts for inclusion/exclusion based on PICO-defined criteria. Reviewers confirmed AI recommendations via an intuitive dashboard, maintaining control while cutting first-pass screening time nearly in half
Automated Data Extraction & Summarization:
Structured extraction templates captured endpoints, populations, and outcomes with automated citation tagging. The AI then generated draft summaries for disease burden, clinical efficacy, and safety profiles which experts refined for final inclusion
Audit & Transparency Layer:
Every generated section contained an audit trail linking AI-generated outputs to their source publications, aligning with HTA documentation standards
The Outcomes
Accelerated SLR Workflow
Introduced structured authoring and semi-automated extraction, cutting overall SLR creation time by 30–40% while maintaining scientific rigor
Streamlined Evidence Screening
AI-assisted triage reduced manual abstract screening by ~50%, allowing teams to focus on high-value clinical insights
Quality Validation Framework
Built SME-led validation checkpoints at each stage to ensure accuracy, compliance, and therapeutic relevance
Faster Market Access Readiness
Enabled payer-ready GVD drafts to be delivered weeks earlier, supporting timely HTA submissions and faster access decisions
.webp)
