Development of a Model for Deconvolution of Heterogeneous Patient Data for Disease Modeling
The emergence of Big Data technologies in the past decade transformed the approaches in healthcare sector. The healthcare industry has started generating and utilizing large amounts of data to predict epidemics, cure disease, improve quality of life and avoid preventable deaths. However, due to the unstructured nature and heterogeneity of healthcare data, using conventional big data-driven models has not been satisfactory. Hence, the need for new knowledge based mathematical methods to address these challenges is necessary and inevitable.
In this project we focus on two different multi-scale clinical applications. In the first research we aim to develop a predictive model for early detection of Sepsis in ICU-patient using multiple data types (Vital signs, Lab results, Medication, Demographic data, …). Other study involves integration of ECG-based parameters, glucose data and dialysis time of a small group of diabetic patients with chronic kidney disease in order to find an association between arrhythmia and glucose level.