Fig. 3

Cultures can be clustered and classified based on their morphofunctional signature. a Hierarchical clustering based on z-scores of all descriptors. The correlation of the descriptors with culture age (DIV) is indicated (black/orange color coded) as well as the descriptor set to which they belong; b PCA based on morphological descriptors distinguished hippocampal from cortical cultures. In a biplot of the first two principal components, data points cluster according to culture time and follow distinct temporal trajectories along the direction of decreasing nuclear count and increasing dendrite- and synapse density (n_bio = 3 x n_tech = 6). A biplot of PCA on functional descriptors could only differentiate early (3–10 DIV) from later time points (14–18 DIV) (n_bio = 3 x n_tech = 6) (Ellipses show the 67% confidence interval); c Confusion matrices and classification results using a random forest classifier. The misclassification rate (MCR) of the classification based on morphological descriptors was 5% (n_bio = 3 x n_tech = 6) and much lower than the MCR of the classification based on functional data (45%) (n_bio = 3 x n_tech = 6); d The z-scores of a subset of descriptors were multiplied with their respective correlation with culture age (weighted z-scores) and summed to obtain a connectivity score that allows facile interpretation of the degree of connectivity at a given time point (n_bio = 3 x n_tech = 6)