Project 11

Multi-parameter vascular assessment for cardiovascular events prediction: from a machine learning approach to a tool ready for clinical practice


In the last years, increasing evidence demonstrated the predictive value for CV risk of a number of biomarkers of early vascular aging. Carotid ultrasound allows having at the same time information about atherosclerosis (carotid intima-media thickness and plaques) and arteriosclerosis (carotid distensibility). Furthermore, radiomics techniques are potentially able to extract a large number of novel image descriptors, exploring wall ultrastructure in an unprecedented way. The integrated predictive value of these biomarkers, obtainable with a single examination, widely performed in routine clinical practice, has never been determined. Aim of the study is to develop in collaboration with a multi-parameter algorithm for vascular aging assessment, based on carotid ultrasound and ready for the general practitioner, able to provide CV risk prediction in a simple, fast and accurate way.

This study will include about 19,000 deeply phenotyped individuals, with state-of-the-art vascular characterization, from the Paris Prospective Study 3 (PPS3) and Maastricht Study cohorts. There is also the opportunity to further enlarge the study population through the activities of the COST Action VascAgeNet. A large number of radiomics descriptors will be extracted from the radiofrequency signals, allowing a high-resolution characterisation of the carotid vascular wall. The ESR will develop machine learning algorithms that will classify individuals according to risk of cardiovascular events, based on arterial structural and functional properties. The PPS3 cohort at Université de Paris will facilitate a training data set (n=10,000), while test data will be available from the Maastricht Study cohort at UM (n=9,000). Secondments at Quipu and at TMC will provide state-of-the-art support about echographic image processing and machine learning approaches.

Requirements:

  • Qualifications: Degree in Engineering, Bioinformatics, Biostatistics, Medicine
  • Experience: Previous experience in biomedical research is appreciated
  • Knowledge & skills: Image processing skills are welcome
  • Attitude and disposition: Willingness to work in multidisciplinary teams