Advancing gut health with NEC predictive modelling
NEC Laboratories Europe is NEC’s European technology research and development center and a world leader in core and applied research in information and communications technology and digital health. Part of seven global interconnected NEC research and development centers, NEC Laboratories Europe is committed to advancing open innovation, reducing carbon emissions and creating a better society in which everyone can reach their full potential.
In recent years NEC has established itself as a leading innovator in digital heath, addressing a wide range of medical challenges, including disease prevention, diagnosis and patient treatment. NEC is a forerunner in the use of computational pipelines that combine bioinformatics and machine learning to develop personalized cancer vaccines. Its biomedical partners include BostonGene, who provides molecular and immune profiling, and Transgene, who used NEC technology to move its head and neck therapeutic cancer vaccine into clinical trials.
NEC Laboratories Europe contributes significantly to NEC’s digital heath portfolio, focusing on graph-based AI methods (GraphAI) and other AI domains, such as explainable AI and computational science, to advance its research. NEC Laboratory Europe’s GraphAI technology integrates multi-modal data to reveal hidden relationships in underlying data, including that of the human microbiome.
ENDOTARGET is the most extensive project of its kind to study the significance of gut microbiota as a driver of chronic systemic inflammation and its role in the pathogenesis of rheumatic diseases. Gut microbiome – the bacteria, viruses and fungi colonizing our gut – is associated with the development of various chronic diseases, including metabolic disease, gastrointestinal disorders and colorectal cancer. However, most mechanisms linking the composition of a patient’s gut microbiome to the development of such diseases are still to be unraveled.
In recent years, there has been growing evidence that endotoxemia – endotoxins released by bacteria that cross the gut barrier and enter the bloodstream – can cause rheumatoid arthritis. Because the gut microbiome plays an important role in maintaining the gastrointestinal barrier, it is hypothesized to be a factor in the development of several rheumatic diseases.
To help specify associations related to imbalances in the gut microbiome, NEC Laboratories Europe developed Multimodal Variational Information Bottleneck (MVIB). This novel deep-learning model can be used to predict whether individuals may be affected by certain chronic diseases like rheumatoid arthritis and represents a significant advancement in the science of gut health. MVIB works by jointly analyzing an individual’s gut microbial species-relative abundances and strain-level markers. The model uses a saliency technique to interpret the output of MVIB and identify the most relevant microbial species and strain-level markers to model predictions.
In ENDOTARGET, its partners utilize their domain-specific knowledge to train an AI model that will predict the risk of developing the rheumatic diseases osteoarthritis, rheumatoid arthritis and spondylarthritis. NEC Laboratories Europe uses MVIB to perform risk prediction. The Swiss Institute of Bioinformatics is leading a work package to identify candidate biomarkers for healthy microbiome and non-healthy (disease transition) microbiome. The University of Tartu is leading the identification of risk prediction scores, which will be used along with microbiome sequencing data.
Beyond the identification of markers and risk prediction, another aspect of NEC’s contribution will be to help identify personalized interventions (such as a more favorable diet). These will help reduce the disease activity in rheumatoid arthritis and spondylarthritis and slow the cartilage damage and progression of osteoarthritis.
ENDOTARGET has the potential to improve the quality of life of patients who are affected by – or have a substantial risk of developing – rheumatic diseases and help reduce the volume of patient treatment hospitals must deliver and manage.