“
TL;DR
A neural network classifier was applied to DNA methylation data from peripheral blood to distinguish control, pre-hypertensive, and hypertensive individuals.
Journal of Personalized Medicine · 2 authors, 1 centre
This summary was generated by AI from a single paper. It has not been reviewed by a clinician and is not clinical advice. Verify against the source before acting on it.
A neural network classifier was applied to DNA methylation data from peripheral blood to distinguish control, pre-hypertensive, and hypertensive individuals.
This study applied artificial neural networks to DNA methylation data from human peripheral blood to classify individuals as control, pre-hypertensive, or hypertensive. A CpG subset selection optimization approach was used to reduce input dimensionality and improve classifier performance. The model achieved 86% mean accuracy distinguishing control from hypertensive/pre-hypertensive patients using 2239 CpGs, and a comparable 83% accuracy using only 22 CpGs. For distinguishing hypertensive from pre-hypertensive patients, 1120 CpGs yielded a mean accuracy of 88.3%. The authors note that using the full set of available CpGs did not generate accurate classifications. The authors suggest that peripheral blood methylation testing could serve as an objective, stress-independent screening tool, though the clinical utility remains to be validated.