New Platform Leverages Real-World Evidence for Type 2 Inflammation Immunological Disease Research

Researchers recently evaluated Immunolab, a web-based interface that generates real-world evidence (RWE) to support research on immunological diseases related to type 2 inflammation, and found that it may be a user-friendly and comprehensive tool for generating insights and analyzing health care data. Researchers published the findings in the Frontiers in Allergy.  

“Real-world evidence has traditionally been used for regulatory and payer purposes, but its use in regulatory decision-making has been increasing,” said researchers. “However, fragmented and inconsistent RWE data sources pose challenges.”

In response, the Immunolab platform was developed to leverage large patient cohorts and to generate data-driven insights and research in immunological diseases. Immunolab was designed to address research questions related to drug development, as well as pre- and post-launch evidence generation needs, said researchers. These insights were projected to be especially impactful for research around type 2 inflammatory diseases with high prevalence, high rates of comorbidity, and diverse clinical management. 

A multidisciplinary team of researchers, clinicians, medical professionals, and health economics experts found gaps in real-world evidence for immunology research and formulated research questions to address these gaps. The team also recognized the importance of a platform solution and distributed access to hypothesis-generation tools. The Immunolab core development and analytic design teams refined cohort definitions and feature designs using real-world evidence to evaluate various scenarios and the impact of key design choices. All decisions made by the teams were recorded in a decision log, which serves as a knowledge repository and aids in the ongoing maintenance of Immunolab.

The platform used patient data from the Optum de-identified EHR dataset culled from over 17 million patients. It included information on 33 type 2 immunological indications and associated comorbidities.

Immunolab was built in a cloud-based system based on a high-performance computing spark cluster, machine learning (ML) libraries, and data visualization tools, said researchers. The interface used maps, drop-down menus, and intuitive graphs to perform and display analyses. It offered 3 analytical modules that use machine learning algorithms to generate results for further analyses: the Patient Journey Mapper (PJM), the Switch Modeler (SM), and the Head-to-Head simulator (H2H). 

These modules provided tools for hypothesis generation and analysis of treatment patterns, treatment switching patterns, and comparative effectiveness of treatments based on clinical outcomes. The PJM module allowed users to explore the characteristics and treatment journeys of patients, while the SM module identified the key factors driving treatment switching. The H2H simulator enabled robust comparisons of different drug classes and adjusts for confounding effects. 

Researchers estimated the platform could facilitate over 7 million rapid “insight generation” analyses. The PJM could provide about 5 million analyses across 70 predefined patient subpopulations. For every switch/augmentation event, the SM could generate nearly 130 descriptive statistics and up to 2 million analyses. The H2H could execute approximately 75,000 descriptive analyses and across 4 therapeutic groups can generate up to 150 descriptive statistics for approximately 150 patients subpopulations. 

“Future expansions integrating new analytic modules and additional data sources into platforms such as Immunolab can put the means of rapid analytic exploration into the hands of researchers, making both data and analytics available to them,” said researchers.

Reference: 
Hamelin B, Rowe P, Molony C, et all. Immunolab: combining targeted real-world data with advanced analytics to generate evidence at scale in immunology. Front Allergy. 2022;3(Therapies and Therapeutic Targets). doi.org/10.3389/falgy.2022.951795