One Literature
Evidence Platform:
No Literature Evidence Silos

Evidence management is a strategic imperative supported by governance, standards and a software solution acting as the workflow engine and centralized evidence repository. The literature review software lives at the core of an evidence management platform. It ingests literature and provides a configurable workflow to evaluate, appraise, extract, standardize and store evidence, which can be easily shared across the organization.     

Literature Evidence is defined as the appraised, validated and standardized facts extracted from a comprehensive body of scientific literature in the course of an evidence synthesis project. Literature evidence is further enriched with an audit trail of who, when and in what context the evidence was extracted, providing the transparency required in health sciences, academia and the public sector. Large evidence synthesis projects can produce more than 100,000 data points.

Key Characteristics of Literature Evidence:

1. Production: Generated through the systematic process of an evidence synthesis project, where facts are extracted and their corresponding context is captured.
   
2. Source: Derived from relevant and graded scientific references which serve as the source of factual information.
   
3. Validated: It is validated as part of the process, establishing the facts were interpreted and extracted properly.
4. Auditable: The entire research process is documented, transparent, and detailed so third parties can retrace every step taken by the reviewers and verify the findings.
5. Standardized: The facts and their context are standardized to ensure uniform data definition and consistency across an organization or field, enabling unambiguous interpretation and use.

DistillerSR replaces manual, spreadsheet-based processes with secure, AI-enabled workflows designed to accelerate every stage of the literature review lifecycle while being regulatory compliant.

Human Oversight: Every critical AI-extracted data element requires human validation (accept/reject) to ensure regulatory compliance with bodies like NICE, FDA and Canada’s Drug Agency.

Compliance, Security and Privacy: DistillerSR has adopted the NIST AI Risk Management Framework, enabling us to govern the design, deployment, testing, verification, and validation of AI capabilities on our platform. All data remains within the DistillerSR environment and no customer data is used to train external AI models. 

Intelligent Screening and Prioritization

AI Rerank: Learns from your team’s ongoing inclusion and exclusion decisions in real-time. It continuously reprioritizes unscreened references, pushing the most relevant studies to the top of the reviewing queue.

AI Classifiers: Works alongside human reviewers to automatically label references, categorizing them by study design, specific intervention, or PICO element.

AI Error Checking: Flags potential human screening errors by identifying excluded references that closely match your project’s inclusion patterns.

Smart Evidence Extraction (SEE)

Purpose-Built GenAI: Unlike general-purpose AI, SEE is built specifically for the unique demands of literature review data extraction. The model is strictly grounded to your source documents.

Automated Suggestions: SEE finds, suggests, and extracts data from scientific journals to answer questions, including numerical content, sentiment analysis, risk of bias information, and study summaries.

100% Traceability: Every suggested data point is directly linked back to its origin in the source document. 

Human-in-the-Loop: Built into DistillerSR’s workflow to ensure users have complete control over accepting the suggested answers to ensure regulatory compliance.

Copyright and Privacy Compliant: All customer data and references remain within the platform and are not shared with third-party GenAI vendors.

DistillerSR Agentic AI

Agentic AI: Combines Large Language Models (LLMs) with your repository of expertly validated evidence to help draft regulatory reports and synthesize insights faster.

Breaking Data Silos: Share validated evidence across your organization, ensuring different functions—from Clinical to Medical Affairs—operate from a single “Source of Truth”.

DistillerSR’s approach to Generative AI is built on a foundation of security, control, and compliance. We recognize that in the field of evidence synthesis, the security of your proprietary data and the copyright of scientific publishers are paramount. Our AI solution is engineered to mitigate the risks typically associated with large language models by prioritizing a “closed-loop” system and strict adherence to publisher agreements.

Core Pillars of our Copyright and Data Security Framework

Data Isolation: DistillerSR AI models are hosted entirely within our secure ecosystem. Your proprietary data and references are never shared with external systems or third-party AI providers, ensuring your research remains private and protected.

Customer-Led Processing Controls: To respect the agreements between research organizations and publishers, DistillerSR provides customers with full control over which references can be processed by AI. This allows you to exclude specific content based on your unique licensing terms.

No Training on Your References: We do not use customer-uploaded references or extracted data to train or refine our AI models. Your intellectual property and the copyrighted materials you process remain solely yours and are never incorporated into the underlying model’s knowledge base.

Aggressive Grounding and Attribution: To prevent “hallucinations” and ensure transparency, our AI utilizes aggressive grounding and pre-processing. Every suggested answer and piece of evidence is attributed directly to the specific source reference, providing a clear and defensible audit trail.

Strategic Partnerships for Compliance

DistillerSR works with global information providers to ensure a seamless, compliant workflow. For example, the Embase digital-based license (DaaS) allows organizations to fully leverage our AI capabilities for evidence synthesis while remaining 100% copyright compliant with Elsevier’s content.