Fuzzy clustering of vitamin and homocysteine levels in patients with history of ischemic stroke

Abstract

Vitamin, metabolite, and homocysteine levels were measured in 166 patients with a history of ischemic stroke. An interactive relationship between vitamin, metabolite, and homocysteine levels was sought. The data analysis used the fuzzy c-means algorithm and fuzzy conditional clustering. The number of clusters in each case was determined by the Xie-Beni Index. In conditional fuzzy clustering, the data was clustered based on the membership value of one of the sets. Different combinations of vitamins, homocysteine, and cysthionine levels were tested. While many patients had low levels of vitamins or high homocysteine, these did not correlate in all instances. Normal vitamin levels were associated with normal homocysteine levels. Homocysteine levels did not appear to depend on vitamin levels alone. Other factors such as vitamin receptor abnormalities may dictate individual needs and ultimately play a role in the interaction between vitamin levels and homocysteine.