NordicRWE was recently awarded a major public research grant from the Research Council of Norway for the period October 2021 to December 2024 (grant no. 327887).

The consortium is led by NordicRWE with partners from the department of biostatistics at the University of Oslo, the department of registry informatics at the Cancer Registry of Norway and NEC Corporation, and their R&D division of AI drug development in Heidelberg, Germany.

In this project, named RWE ACCELERATE, we will during the next three years turn the unique health data into new products and services that will accelerate the drug development process.

The research will target two highly relevant and interesting areas of integrating and applying RWD/E into the drug development and drug cycle management processes:

1. Constructing oncology external control arms (ECAs) to emulate and reproduce results from RCT

The first proof-of-concept step is to emulate the trial populations in already completed RCTs and run analyses with the aim to reproduce the outcomes in the standard-of-care (placebo) and experimental treatment arms. De-identified individual level data from completed RCTs will come from pharma companies directly or via 3rd party platforms. We will explore methodological aspects for ECA, e.g., temporality and intermediary outcomes. In addition, the project will construct synthetic datasets that can be used for feasibility assessment and data exploration.

NordicRWE, is looking for partners from pharma who can attach to this project, e.g., to define concrete use-cases, bring own data, bring know-how and expertise and work together with us on this interesting research.

We have decided to focus on lung, ovarian, colorectal and breast cancer in the first phase, but can expand to other cancers later.

2. Combining network-based machine learning algorithms and pharmacoepidemiology to identify and validate drug re-purposing candidates and side-effect signals in large registry datasets

With this research project we propose a novel data-driven alternative for drug discovery and drug surveillance that combines in silico and machine learning hypothesis generation with subsequent rigorous pharmacoepidemiological hypothesis validation using longitudinal real-world data (RWD).

In brief, the following stepwise approach will be applied, but also re-iterated and refined over time as new knowledge is generated and data accumulates in the registries

i. Firstly, we will tap into publicly available databases in order to generate testable hypotheses using machine learning and network-based algorithms

ii. In parallel, we will use machine learning based algorithms and biostatistical methods to run hypothesis free drug wide association studies (drug-scanning) directly on the high-dimensional real-world data with the aim to detect associations between drug exposure and disease onset or side-effect signals

iii. Next, we will use the information generated from the first two steps to select the most promising candidate drugs and pursue them further as drug re-purposing candidates or likewise, select the side effect signals of interest

iv. Finally, in a hypothesis refinement step, we will conduct hypothesis validation applying rigorous patient-level pharmacoepidemiologic analyses to evaluate the association between exposure to the repurposing candidates and incident disease onset or side-effects of focus.

This program will focus on degenerative disease (e.g. dementia) and cardiovascular drug classes (e.g. lipid-modifying drug), but with the possibility to expand to other phenotypes later.