DC-FMDT: A Fuzzy Model Decision Tree Algorithm Based on Drug Combination Features

With the development of the field of medicine, more and more diseases have been cured. To analyze the relationship between drugs and diseases faster, people began to use machine learning technology as a means to assist doctors in treating diseases. There are hundreds of known methods for medicine, and a large proportion of these techniques use deep neural networks, which are difficult to interpret model structures. This shortcoming is magnified in areas such as medicine, which require ethics because the price of misjudgment is enormous. If the doctor does not know the decision-making logic of the model clearly, it will cause great insecurity. To this end, we propose a fuzzy model tree algorithm DC-FMDT, which combines the characteristics of the pairwise appearance of drugs in the combination drug dataset, which not only preserves the interpretability of the decision tree model but also has good performance. Finally, our algorithm is validated in the drug combination dataset of the two cell lines.