Case Study: Roche Real-World Evidence Integration for Enhanced Benefit-Risk Analysis

Company

Roche / Genentech (Top 5 Global Pharmaceutical)

Therapeutic Areas

Neurology (Ocrevus/ocrelizumab) & Oncology (Tecentriq/atezolizumab)

Focus

Real-World Evidence integration into BRA

Challenge

Unifying disparate RWD sources with clinical trial data

ArcaScience Modules Used

Data Intelligence, Decision Intelligence, Automated Outputs

Timeline

16-week implementation

The Challenge

Fragmented Real-World Data Across Multiple Sources

Roche recognized the strategic value of integrating real-world evidence into their benefit-risk assessments but faced significant technical and methodological challenges. Real-world data was scattered across 12+ sources—claims databases, EHR systems, registries, and patient support programs—with different formats, terminologies, and quality standards. Manual data harmonization took months and produced inconsistent results. Regulatory agencies were increasingly requesting RWE to supplement clinical trial data, but Roche lacked a scalable framework to deliver this systematically across their portfolio.

The Solution

ArcaScience Platform Deployment

ArcaScience deployed its Data Intelligence module to create an automated RWE integration pipeline, harmonizing 12 real-world data sources with clinical trial databases for continuous, enriched benefit-risk analysis.

Implementation Process

Weeks 1–3

Data Source Mapping

Inventoried and assessed all 12 RWD sources. Mapped data elements to common data model (OMOP CDM). Established data quality scoring and validation rules.

Weeks 4–7

Integration Pipeline Build

Built automated ETL pipelines for each data source. Implemented AI-powered terminology harmonization (ICD-10, MedDRA, SNOMED-CT mapping). Created unified patient-level longitudinal records.

Weeks 8–11

Signal Detection & Analysis

Deployed cross-source signal detection algorithms. Identified 23 new safety signals not visible in individual data sources alone. Characterized signal strength using multi-source triangulation methodology.

Weeks 12–14

BRA Enhancement

Integrated RWE findings into existing MCDA benefit-risk frameworks. Generated enhanced PBRER sections incorporating RWE. Created regulatory-ready data packages for health authority requests.

Weeks 15–16

Validation & Handover

Validated results against known safety signals (100% sensitivity). Trained Roche's epidemiology and PV teams. Established ongoing automated monitoring.

Results

75%

Reduction in data
preparation time

sources

Unified real-world
data sources

signals

New safety signals
identified

Additional result: 40% faster PBRER completion, reducing turnaround from 14 weeks to 8.4 weeks.

Quantitative Outcomes

Metric	Before ArcaScience	After ArcaScience
RWD preparation time	4–6 months per analysis	2–4 weeks automated
Data sources integrated	2–3 (manually selected)	12 (comprehensive)
Signal detection from RWD	Ad hoc, retrospective	Continuous, automated
New signals per year from RWE	3–5	23 in first cycle
PBRER completion time	14 weeks	8.4 weeks (40% faster)
RWE regulatory submissions	2 per year	8 per year

"ArcaScience solved a problem we'd been struggling with for years. Integrating real-world evidence into our benefit-risk assessments at scale seemed impossible with manual processes. Their platform made it not just possible but routine—and the signal detection capabilities exceeded our expectations."

— VP, Real-World Evidence & Epidemiology, Roche

"Finding 23 new safety signals from integrated RWD analysis was a breakthrough. Several of these signals informed labeling updates and risk management strategies that we would not have identified from clinical trial data alone. This is the future of pharmacovigilance."

— Global Head of Patient Safety, Roche/Genentech

Key Takeaways

Real-world evidence integration at scale requires automated data harmonization—manual approaches cannot keep pace with regulatory expectations.
Multi-source signal triangulation identifies safety signals invisible to single-source analysis, providing a more complete safety picture.
The 75% reduction in data preparation time transforms RWE from a special project into routine PV operations.
Regulatory agencies increasingly value RWE-enriched BRAs, making scalable integration a competitive necessity.
Continuous automated monitoring replaces periodic retrospective analyses, enabling proactive rather than reactive safety management.

Ready to Transform Your Benefit-Risk Strategy?

Learn how ArcaScience can support your next regulatory submission with AI-driven benefit-risk analysis.

Schedule a Consultation | info@arcascience.ai

Roche: Real-World Evidence Integration for Enhanced Benefit-Risk Analysis