Abstract: A balanced approach to processing medical data from studies of mental illnesses for the elderly people is proposed. The task is far from being solved and rests on three fundamental problems:• The set of laboratory research data is typically insufficiently complete and contains many unfilled fields;• It turns out that the appearance of unfilled fields or the method of filling them greatly affects the final result;• It turns out that the use of data from a limited number of laboratory tests does not allow for a reliable diagnosis, therefore, for correct diagnosis, it is necessary to first identify significant diagnostic parameters. The paper proposes an approach that allows solving these issues and substantiating a variant of computer diagnostics of diseases in the elderly. According to the International Organization for the Fight against Alzheimer’s Disease, currently around 55 million people worldwide suffer from dementia and the number of such people is projected to only grow. The identification of early signs of incipient dementia remains quite difficult, preclinical and pre-clinical diagnosis of cognitive impairment is significantly complicated by the slow, imperceptible progression of cognitive impairment characteristic of neurodegenerative diseases with the development of symptoms, which in the early stages can be regarded as a normal age-related cognitive decline. The use of several examination methods provides a chance to increase the accuracy of early diagnosis of Alzheimer’s disease. However, modern methods of computer analysis (spectral, autocorrelation, cross-correlation, the method of complex mutual spectra, etc.), methods of functional neuroimaging (single-photon emission tomography (SPECT), positron emission tomography (PET), etc.), magnetic resonance imaging (MRI) used for diagnosis reveal such a large heterogeneity of patient examination results, what becomes clear is that none of them can be so precise and specific as to be the basis for a diagnosis of Alzheimer’s disease. The paper proposes an approach based on modern consolidation methods for processing medical examination data of patients. © 2025 Elsevier B.V., All rights reserved.