Transcriptomic profiling and machine learning models predict cerebral small vessel disease and dementia

Machine learning models using transcriptome and neuroimaging data predicted cerebral small vessel disease (average F1=0.93) and dementia (F1=0.73) in 91 patients. Differentially distributed neuroimaging features included white matter hyperintensity severity and CSVD burden.

• Dementia vs. non-dementia subgroups showed significant differences in total white matter hyperintensity severity, CSVD burden, neutrophil counts, and triglyceride levels.
• Transcriptome-wide analysis identified several potential blood-based biomarkers for CSVD, validated by qPCR.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.