Technology unit
Data Science, Data Management and Digital Solutions

We have shaped our data science core to respond to the wide range of outcomes that might be expected in a multi-dimensional project.
The following list describes most of our capabilities, in order of complexity:

• we build in silico models to assist with decision-making,

• we build both predictive and explicative models to decipher molecular signatures (identify biomarkers) and mechanisms of action (MoA), and to provide complete functional analyses, including differential analysis, pathway enrichment and biological networks inferences (for example, using BTM, and WGCNA analysis),

• we extend the biological relevance using existing reference databases or those relating to other species,

• we provide a high level of integration of multi-Omics, cytometric, and unstructured readout (e.g. imaging) with the results of (pre)clinical studies and other real-world data (RWD),

• we envision multi-scale analyses and event correlations by incorporating spatial (e.g. local vs. systemic) or temporal (e.g. longitudinal studies) dimensions,

We generate most of these biological outcomes through the intensive use of a combination of advanced statistical, mathematical, and machine learning (ML) approaches. Our team of data scientists has demonstrated its expertise in the benchmarking and application of such approaches in both unsupervised and supervised integrative analysis.

To be able to better integrate unstructured readouts, such as images, in intensive analysis or to address new decision making needs, such as faster/better human-mimicking, we are adding to our capabilities using recent deep-learning approaches and neural network algorithms; for example, CNN applied to Computer Vision through TensorFlow frameworks.

• Application of AI/ML: Holistic analytical approach

Data tracking, security, and integrity throughout the data life cycle are supported by digital solutions. The portfolio of solutions that we can use in your project comprises:

• generic platforms for data warehousing and data sharing in a big data context,

• custom solutions for the specific characteristics of the harvested data; e.g., LIMS, eCRF, LabKey (centralized multi-study projects), and tranSMART (clinical hypothesis testing)

We guarantee you a high level of Data Governance and the application of the required standards through:

• the collaboration of our data steward, an internal multi-business executive committee, and data managers,

• the ontology-driven organization and description of data/metadata,

• the application of FAIR principles (Findable, Accessible, Interoperable, Reusable).

Such platforms reduce data complexity, facilitate the presentation of complex concepts, offer magnified or user-context knowledge representation, and offer business-focused interfaces. This allows a final user (a scientist, clinician, or patient) who is not familiar with bioinformatics, statistics, or digital matters, to interact with the data and fully understand the data.
We can design a broad range of applications, according to your needs:

• decision-making/support platforms; for example, in a clinical context, using companion software,

• hypothesis generation, testing, and/or validation; for example, in clinical context, using (tranSMART, etc.),

• “DataMining4DataMeaning”: mined data is used to construct and assess predictive models,

• knowledge aggregation, supported by complex querying tools,

• registration and integration of (meta)data to facilitate advanced data management; for example, using LabKey.

We suggest the most appropriate technological approach for your project, through:

• customization of off-the-shelf digital solutions,

• use of Web-applications, such as Rshiny and notebooks,

• design, development, and deployment (DevOps, Continuous Integration, etc.) of specific applications by instantiating methods and tools at the core of our Software Engineering Platform.

The conformity of our information systems with security standards is a daily preoccupation, in order to guarantee data integrity and data privacy. We use monitoring and supervision IT business standards, embedded in certified ISO9001 and internal ISS Policy frameworks, and assessed by self-or partner-mandated audits.
These data privacy and cybersecurity concerns are of great importance with regard to compliance with GDPR requirements. We have demonstrated such compliance in clinical study-related privacy impact assessments (PIA), performed as controller or joint controller.