*Purpose: KTD-Innov is a French consortium launched in 2018 and dedicated to improving the personalized follow-up and outcomes prediction for kidney transplanted patients. In addition, the French historical DIVAT database for kidney transplantation has collected demographic and clinical data for the recipient and the donor, graft characteristics, treatments and follow-up data for more than 8,000 patients. We ambition to integrate these data into a personalized medicine application to help clinicians taking clinical management decision for their patients. We have hence developed contextualization algorithms to project a patient of interest (POI) within a population of similar patients (POR, population of reference) in order to better describe, manage and anticipate post-transplantation outcomes.
*Methods: We first selected 23 relevant variables, according to clinicians’ expertise, including donor and recipient’s age and body mass index. We proposed 3 personalized POR selection with contextualization algorithms. (1) The filtering approach: the clinician can select the values or the scale of values for each variable to select the POR. (2,3) The distance approaches, based on a dimension reduction method: the user selects (2) the number of nearest neighbors or (3) a pre-calculated cluster, from hierarchical clustering analysis.
*Results: After the selection of POR, we compare their biological values during the follow-up visits to those of the POI. Growth-like curves are calculated and displayed to represent kidney function-specific biological levels (e.g. creatinine) of a POR during follow-up and thus project the POI evolution in this space. We built a clinical decision support system, called iDS, to integrate these 3 contextualization algorithms and offer a personalized monitoring and follow-up of individual trajectories of post-kidney transplantation outcomes.
*Conclusions: This application facilitates access to large amounts of data by allowing their visualization and comparison as a way to optimize medical care and guide clinical decisions.

Olivia Rousseau

Nantes Université, CHU Nantes, Centrale Nantes, Inserm, Center for Research in Transplantation and Translational Immunology, UMR1064

Appearances