Prospective immunological markers of cerebral microangiopathy and Alzheimer's disease

Evaluate the cognitive and immune status of patients with cerebral microangiopathy and Alzheimer's disease. 85 patients with cognitive decline according to the Montreal Cognitive Assessment Scale were examined and divided into three groups: Group 1 included 25 patients with cerebral microangiopathy; Group 2 included 15 patients with Alzheimer's disease; Group 3 included 25 patients with mixed dementia. The control group consisted of 20 age-matched volunteers without any brain lesions. Cognitive status was assessed using Montreal Cognitive Assessment Scale and tests to evaluate praxis, gnosis, regulatory and neurodynamic functions. Laboratory diagnostics included assessment of interleukin concentrations, IFN-g, TNF alpha\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha$$\end{document}, GM-CSF, CC-chemokines. Statistical methods and machine learning algorithms were applied. In patients with cerebral microangiopathy, the first place was occupied by disorders of executive functions, while in patients with AD-by amnestic, semantic and perceptual disorders. Evaluation of serum concentrations of cytokines of different groups revealed lower levels of GM-CSF and TECK/CCL25 in patients of groups number 2 and 3, higher levels of IL-6, IL-1b, IFN-g and TNF alpha\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha$$\end{document} in patients of group number 1, expression of MIP-1a/CCL3, MPIF-1/CCL23 and MIP-3a/CCL20 in patients of the main groups. Machine learning methods (Decision Tree) trained to separate patients from different groups did not score above 0.75 (f1 score). This step was made to bring current medicine closer to AI-enabled personalised medicine. The results of the present study demonstrate promising directions in the search for new immunological markers of the pathogenesis of major conditions accompanied by cognitive dysfunction. However, machine learning models showed low results. Therefore, results obtained indicate the need for further multicentral studies with a large sample size to determine the potential value of cytokines of different groups as immunological markers of cognitive disorders of various etiologies.