© 2021Medical diagnosis is a disease identification process that matches symptoms with diseases based on the symptoms of target patient. In this process, it is necessary to establish a similarity relation between symptoms and diseases so as to determine the correct diagnosis. Similarity measure theory is a beneficial way that is used to model this relationship mathematically under vary environment. In the literature, various similarity measures have been constructed in single-valued neutrosophic set setting. However, these similarity measures ignores the interaction between symptoms. To overcome this deficiency, we propose four new similarity measures by using the Choquet integral under single-valued neutrosophic environment that take into account both period and the interaction between symptoms. Moreover, we take advantage of the concept of 2-additivity to reduce the computational effort to obtain multi-period medical diagnosis results. We implement them to a multi-period medical diagnosis example existing in the literature. We also compare our results with some previous ones and we analyze the consistency of the results via some statistical methods.