Background:

To investigate the resilience of the autonomous nerve system (ANS) a resilience experiment, with five different stages (mental, hand grip, supine, standing, and walking), was executed in the third-generation participants of the Framingham Heart Study (around 1300 participants). Several vital signs of the heart (systolic blood pressure, diastolic blood pressure, mean arterial pressure, heart rate, and respiratory rate) were continuously observed throughout the experiment (25–30 minutes of monitoring). The experiment is visualized in the following figure.


Data Analytic Questions:

This type of data is not easy to analyze due to the large number of repeats and its high correlation among sequential data points. One example of the data that is generated is provided in the figure below.


Each person will have their own vital sign profile and all participants react differently to the different stages. Additionally, we have already information on sex and age on each participant, but we are currently trying to add other information (e.g., BMI, smoking, hypertension, past heart problems, treatments) from the participants. 

For the master thesis project, we are trying to answer several data analytical questions:

·        Can we describe the longitudinal vital sign signals properly (functional data analysis, time-series analysis, longitudinal smoothing) and find clusters of participants?

·        Can we define important features calculated per stage and rest period (e.g., average, variation, delta response, autocorrelation, time at maximum response) that summarize the most relevant information in the trajectories?

·        Can we determine the association between certain participant characteristic (age, sex, smoking, etc.) and the vital sign profiles (using the profiles or the features).

Together with the Department of Biomedical Engineering (Computational  Biology), I am offering a Master's project that can be started in  January or later.

Suppose we are given a set of patients and for each patient the same  medical measurements have been made (measuring bio transmitters etc).  The goal of computational biology is to inspect the measurements and to  detect rules between different transmitters that explain in which  situations the human body produces certain proteins (which might be  relevant in cancer diagnosis). Having such rules, one can develop  medical treatments that prevent the body to produce these proteins. To  be able to develop effective medical treatments, one is looking for  simple rules between different bio transmitters e.g., transmitters 1 and 2 are active and transmitter 3 is inactive.

The goal of this Master's project is to develop and implement integer  programming techniques that allow to find rules between different bio  transmitters. Among others, we expect that techniques such as column  generation and branch-and-price can be useful to achieve the desired  goal. Since implementation is an important aspect of this project, good  programming skills in Python and/or C/C++ are required.

Are you interested in this project or do you want to know more about it,  please contact Christopher Hojny (c.hojny@tue.nl).