Fig. 4
From: Cardiac tissue engineering: state-of-the-art methods and outlook
Machine learning for drug screening on human iPSCs-derived engineered cardiac tissue. a Waveform pattern parameters are determined based on concentration of cardioactive compounds compared to the binary support vector machine (SVM). The collected data points would be in line with those of vehicle as if the compound does not modulate the contractile behavior of human ventricular cardiac tissue strips (hvCTSs). If data of cardio active effects are more distinguishable, it shows in a higher SVM accuracy which is possible to separate two compound groups. The degree of cardio activity of a given concentration for target compound is shown in a singular quantitative index with the binary SVM approach. b Library of compounds is built on a model for prediction of mechanistic action of screened compounds. Data from the library group allow the machine learning defines boundaries of various drug families. Finally, the developed model can be applied for the unknown compounds on tissue engineering. The image is reproduced with permission from [41]