For each concept of my dataset I have stored the corresponding wikipedia categories. For example, consider the following 5 concepts and their corresponding wikipedia categories.
The question appears a little unclear to me and does not seem like a straightforward problem to solve and may require some NLP model. Also,the words concept and categories are interchangeably used. What I understand is that the concepts such as enzyme inhibitor, bypass surgery and hypertriglyceridimia need to be combined together as medical and the rest as non medical. This problem will require more data than just the category names. A corpus is required to train an LDA model(for instance) where the entire text information is fed to the algorithm and it returns the most likely topics for each of the concepts.
https://www.analyticsvidhya.com/blog/2018/10/stepwise-guide-topic-modeling-latent-semantic-analysis/