Operational models to access multi-modal real-world data in the Swiss and European ecosystem to enable machine-learning derived insights in drug development

Partners needed:

From the academy and public.

• • Hospitals, representatives of consortia, and other data owners to explore and develop privacy-preserving solutions as well as collaborative operational models that enable use of their data for developing tomorrow’s therapeutics.
  • Researchers who study and apply concepts for privacy-preserving analyses on real-world patient-level data.

Idea:

We envision the use of multi-modal real-world data and machine-learning derived insights to enable biotech and pharmaceutical companies to develop the right therapeutic for the right patient population. This will enable reduced timelines in preclinical development and lower failure rates in clinical development, which will increase efficiency and lower costs in drug development.

With this call for collaboration, we aim to explore and develop privacy-preserving operational models that enable the use of Swiss and European patient data to inform drug development. This will bring more diversity in data-driven decisions in drug development and benefit patient groups which are currently underrepresented in the US-dominated real-world data landscape.

We have multiple years of experience in the pharmaceuticals industry, leveraging multi-modal real-world patient data to inform decisions in early drug development. Having seen that most of the data being used is exclusively from the US and often originating from the same geographies with a high density of academic research centers, we see a clear need to improve diversity in data sources.

Our company is building a tech-stack that includes the following steps to deliver insights for drug development.

• • Data ingestion: Identification and curation of suitable data from public and proprietary sources.
  • Deep phenotyping: Deriving features from structured clinical and unstructured data (histopathology images, multi-omics, medical imaging etc.) using a blend of machine learning, epidemiological, and biostatistical methods.
  • Outcomes analysis: Identifying patients with unmet medical need and patient groups with expected high drug effect to provide the best indications, patient groups, and molecular targets for the development of novel therapeutics.

We want to embed all the above in a privacy-preserving framework for data access. To this end, we are looking for collaborators to assess the feasibility of using multi-modal real-world patient data in a European privacy framework and to come up with collaboration models that are attractive to data owners.

Objectives of this call:

• Assess and evaluate different operational models and incentives for integrating Swiss and European data in data-driven decisions for drug development.
• Assess and discuss different technical models for decentralized analysis, such as federated learning or use of synthetic data.
• Map out data owners and stakeholders in Switzerland and how they can be included in a framework for data sharing.
• Build partnerships with data owners.

—–

If you are interested in participating, please reply by May 31, 2024.

If you are interested and want to contribute with your organization, send us the feedback form attached to databooster@data-innovation.org until May 31, 2024, stating

• What is the competency that you could bring in?
• Do you have specific experience that might be relevant in the project context?
• What is the contribution to the project goals that you want to bring in?

The next steps will be:

• Based on your proposed contributions, the call owner will decide about the partners to continue on shaping the idea.
• Afterward: an additional workshop or/and definition of the concrete implementation plan.

The rules of the game: Decisions on whom to invite for the first meeting, and whom to select for the workshops and final innovation team will be made by the company, based on the provided information.

The goal is to set up an optimal innovation team for reaching the goals, not to create a team with as many partners as possible.