Biomedical data analysis using neuro-fuzzy model with post-feature reduction

Now-a-days, a large volume of biomedical data are continuously generated from various biomedical devices and experiments due to the rapid technological advancement in medical science. The effective analysis of these biomedical data such as extracting the significant features biologically and diagnostically is really a challenging task. This paper proposes a Neuro-Fuzzy model with post-feature reduction to analyze these complex biomedical data. The proposed Neuro-Fuzzy approach uses class belongingness fuzzification of input patterns to handle uncertainty issues. However, the complexity of the model increases due to this fuzzy expansion of input patterns. On the other hand, all the expanded fuzzified patterns may not always be significant for model identification. To address this issue, post-feature reduction has been employed on fuzzified patterns to filter out the irrelevant, redundant and noisy features. Unlike pre-feature reduction, this allows all the features to be participated in the fuzzification process and then identify irrelevant features from the fuzzified patterns. Further, this approach allows exploring potential fuzzified features from the strong as well as weak feature set. The effectiveness of this proposed model has been tested and validated with a variety of benchmark biomedical data collected from various domains. (C) 2020 The Authors. Published by Elsevier B.V. on behalf of King Saud University.

Automated summary

This paper proposes a Neuro-Fuzzy model with post-feature reduction for analyzing complex biomedical data. The model handles uncertainty by fuzzifying input patterns and employs post-feature reduction to filter out irrelevant, redundant, and noisy features. The effectiveness of the proposed model has been tested and validated with benchmark biomedical data.