optimization: data quality metrics, subtractive clustring, dynamic ANFIS

I have been researching on data quality metrics and have developed a fuzzy inference system based on probability. The initial FIS was developed with a subtractive grouping algorithm and I used ANFIS for tuning purposes of the membership function. The problem I have is that if my dataset changes over time, for example some of my data points expire and some of them are modified and some of them are even added to my dataset I will have to change the number of groups using adequate radio influence. How can I do a dynamic clustring subtraction algorithm? What I'm asking isn't specifically related to the subtractive substrate algorithm. I need to dynamically monitor my dataset and realize that it is changing (data modification, data expiration, data inclusion). How is this possible over time (any algorithm or approach that can help me).
Thank you