Fig. 3
The decision tree model built for predictive ranking of the attributes relevant to assessment of AgNPs-induced toxicity. This learning model was created from a small database consolidating the results of cell viability analyses described in this work. The first decision tree A contains four parameters: exposure dose, cell type, AgNP type (SCS, LCS, SAS, and LAS), and exposure time (24 and 48 h) and the second one B contains five parameters: exposure dose, cell type, particle size (larger-sized v.s. smaller-sized), surface coatings (citrate v.s. cysteamine). The outcome at each branch terminal is either “nontoxic” (NT-the white square) or “toxic” (T-the gray square), and the numerical data given below the outcome (NT or T) is in the form of n1/n2, where n1 represents the total number of data results (NT or T) and n2 represents the number of data results incapable of fulfilling the outcome