You2Yourself – Towards personalized health monitoring #SWI2022
You2Yourself (Y2Y) develops methods to measure biomarker profiles in periodic body fluid samples of individuals and algorithms to enable the early detection of life-threatening diseases in these profiles, which enhances the chance of cure. Next Generation Sequencing in blood and urine samples is used to establish personal baseline profiles at health. This enables sensitive detection of changes that are associated with the onset of disease. The specific biomarkers that change, indicate the nature of the developing disease (e.g. cancer and type). This idea builds on the fact that every individual is unique and so are his/her biomarker profiles and that personalized monitoring will allow more sensitive detection of disease than current population-based diagnostics.
More information can be found on https://you2yourself.com/.
Y2Y develops statistical methodology and AI algorithms for the detection of cancer and cardiovascular diseases based on periodic personal biomarker profiles. In order to validate its technology, Y2Y is building a unique biobank of sample series that capture the onset of disease, by collecting periodic samples from a large cohort of healthy people, 7% of which will develop disease during the course of participation.
We are looking for efficient algorithms which can detect deviations from a baseline or changes over time in complex, high-dimensional biomarker data. A particular challenge is the longitudinal structure of the data, i.e., several samples of each individual are collected over time. While at the moment, we are looking at individual time points one by one, of particular interest is to incorporate (and profit from) the dynamical structure inherent to the data. Moreover, biomarker profiles for different conditions typically overlap, leading to a blurring of the profiles of the conditions of interest and making detection more difficult. The goal of this project is therefore to develop and implement (one or more) alternative models that make use of the additional structural information. Of particular interest is the extraction (i.e. deconvolution) of single disease profiles.
Y2Y will provide realistic input data for the model. Based on these data, the methodology developed in this project should also be evaluated to obtain quantitative evidence that early detection is improved by longitudinal sampling over single timepoint profiling.