On the separation of normal and abnormal stem cell-derived cardiomyocytes' calcium transient signals

A data set of 438 calcium transient signals measured from induced pluripotent stem cell derived cardiomyocytes was collected to analyze and separate abnormal signals corresponding to aberrant cardiomyocytes from normal signals corresponding to normally developed cells. After the calcium transient peak detection, the authors computed peak variable values. Each signal peak was determined to be either normal or abnormal. The peak variables were used for machine learning algorithms to classify entire calcium transient signals into normal or abnormal types. The authors evaluated the classification power of 10 variables to separate normal signals from abnormal ones. This article obtained classification accuracies of up to 85-95%, around 5% better than the results in the preliminary research. The correctness of the classification of the signals was inferred either by a biotechnology expert or by an algorithm. The new results are promising for the continuity of this area of study in identifying aberrant calcium transients.