Analysis of multi-omics and clinical data: the ICAN I/O platform

Our ICAN I/O platform - Data sciences

In today’s scientific and medical fields, the generation of vast amounts of data, or “Big Data”, necessitates the use of mathematical modeling, statistics, and artificial intelligence. These tools are essential for efficiently managing and analyzing health data, revolutionizing approaches to biology.

ICAN I/O is a platform of expertise focused on multi-omics data analysis and integration, biomarker discovery and clinical research. Our aim is to ensure secure data management throughout its lifecycle, in compliance with RGPD requirements.
The platform goes hand in hand with the evolution of modern technologies and the use of robust algorithms to achieve research objectives through a collaborative and holistic approach integrating multi-omics technologies and statistical analysis.
The team includes data managers and a data scientist who all have extensive experience gained from working on national and international projects, collaborating with eminent scientists, doctors and researchers on many scientific topics.
With this level of expertise, IHU ICAN is recognized as a major international entity capable of generating significant healthcare data, driving advancements in the field.

View charter

Current major projects

MAESTRIA : Machine Learning Artificial Intelligence Early Detection Stroke Atrial Fibrillation

Consortium of 18 partners from Europe, the United States and Canada responding to an H2020 call for projects on digital diagnostics.
Ultra-innovative project to better detect atrial cardiomyopathy, responsible for the occurrence of atrial fibrillation and vascular embolic accidents, with the use of machine learning and AI (Artificial Intelligence).
The ICAN I/O Data Science team plays a central role in integrating multimodal data (imaging, clinical, omics) to improve diagnostic accuracy and identify new therapeutic targets, thereby increasing treatment efficacy and preventing complications.
The ICAN I/O team is also leading the development of the MAESTRIA demonstrator, an innovative digital platform designed to guide atrial fibrillation risk assessment and management, improve diagnostics and support treatment decisions through a scalable, user-centric approach.

Find out more

ARTEMIs : AcceleRating the Translation of virtual twins towards a pErsonalised Management of fatty lIver patients

The ARTEMIs project is developing “virtual twins” using computer models and machine learning, integrated into a Clinical Decision Support System (CDSS) for the management of metabolic fatty liver disease (MFLD).
Targeting a disease that affects 25% of the population, ARTEMIs also focuses on early intervention for the cardiovascular complications associated with MAFLD.
This system aims to personalize treatment by predicting disease progression, assessing treatment efficacy and encouraging a healthier lifestyle.
The project concludes with a proof-of-concept study to evaluate the effectiveness of this integrated approach in a clinical setting.

Find out more

MEDITWIN: using digital twins to develop tomorrow’s personalized medicine

Consortium comprising 7 University Hospital Institutes (IHUs), including IHU ICAN, Nantes University Hospital, Inria, associated startups and Dassault Systèmes.
Project France 2030 announced on Monday December 11, 2023 in the presence of French President Emmanuel Macron.
Using customized virtual twins of organs, metabolism and cancer tumors
5-year term, from 2024 to 2029
Objectives: propose a diagnosis of cardiovascular disease risk for high-risk patients and help specialist physicians choose treatments for effective patient follow-up.

Find out more

RESIST-PP: the link between diet, innate immunity and risk of infection

European Union-funded project to explore how high-fat diets induce postprandial inflammation and atherogenic dyslipidemia via metabolic endotoxemia influenced by gut microbiota.
The project aims to study the impact of bacterial-derived metabolites on the susceptibility of obese and diabetic patients to infections, to analyze interactions between immune cells and blood microbiota, and to carry out studies on mouse models.
The ICAN I/O platform contributes to this project by integrating multimodal data, improving our understanding of diet-induced infection risks, and developing new treatment and prevention strategies for metabolic and infectious diseases.

Find out more

Database management and curation

CRF paper

Protocol reading
Paper CRF review
Paper CRF Validation

eCRF creation

List of variables
Implementation in REDCap
FRC test
Validation of CRF
Creation of user accounts
Release

Edit Checks

Creation and validation of the Data-Validation Plan
Development / Test / Validation – Selection of required fields

Extraction

Data extraction / Creation of datasets

Documentation

REDCap User Guide
Study entry guide
Data-Management Plan of the study

External Data

Link to external data

eCRF Maintenance and Data Quality Control

Fixing problems
New version / Amendment
User Management
Creation of queries

Monitoring

Provision to the monitor of a fraction of data to be monitored

Multi-modal Integration

The ICAN I/O platform specializes in multimodal integration and actively participates in numerous national and European projects focusing on omics studies, working closely with the ICAN Omics Lipidomics and ICAN Omics Metabolomics platforms.
ICAN I/O uses a range of analyses from classical statistical analysis to advanced machine learning algorithms to achieve the study objectives.
The ICAN I/O team is continually developing customized visualization tools to enhance the presentation and interpretation of results.
The platform’s solid experience is backed up by a number of major scientific publications demonstrating its expertise.

Omics

Clinical research

The ICAN I/O platform leverages a comprehensive blend of statistical and machine learning techniques to excel in clinical research.
It crafts Statistical Analysis Plans and performs analyses across Descriptive, Multivariable, and Longitudinal Time-series disciplines using various statistical models The platform’s expertise extends to risk prediction, discriminant analysis, model calibration and validation, as well as advanced classification methods such as the clustering and ROC analysis. It also manages propensity score matching, competitive risk analysis, survival analysis with Kaplan-Meier methods and multiple imputation followed by a sensitivity analysis.
In machine learning, ICAN I/O employs Ensemble Methods, XG Boost, Random Forest, SVM, and PLS-DA, customized to fit specific study requirements and data types It focuses on technologies like Explainable AI, Causal Inference, and Deep Phenotyping, and utilizes Automated ML and Incremental Learning for enhanced data handling.
The platform has developed web applications based on XG Boost for clinical use, discovered prognostic markers and biomarkers, and developed clinical scores, securing its leadership with two international biomarker patents.
His innovative approach has been highlighted by multiple high-impact publications in journals such as The Lancet, European Heart Journal, Nature Cell Biology and others.