Mathematical methods for interpretation of metabolic signals from living cells on biohybrid sensor chips

Within the past few years, sensor chips have been developed that permit monitoring of physiological parameters of living cells - e.g. pH and pO ₂ of the nutrient solution - in real-time. Long term experiments, in which these parameters are screened, produce large amounts of data. Their evaluation is usually carried out by hand in an errorprone and time consuming process. To avoid these problems an automated software solution is needed. Due to the periodic measuring procedure that is common for sensor chip-based systems, the data itself is also periodic. This was taken into account to reduce its size and complexity: by capturing the essential information within each period, time series were computed that offer a more direct insight into the cells' state of vitality. At the same time the data needed for further evaluation was strongly reduced and freed from the influence of sensor drift. An experiment with human breast adenocarcinoma cells of type MCF-7, in which the mentioned parameters were recorded for several days, is used to demonstrate this fact. The impact of a cytostatic drug on the cells' vitality was measured by employing gradient and amplitude analysis as well as other mathematical methods. This process was carried out automatically using the R environment for statistical computing. This approach not only delivers interpreted and therefore more meaningful data. The possibility to automate this process also offers simpler and faster means of evaluating an experiment while helping to avoid human errors. The described methods can be adapted to other sensor chip-based systems.