Mental State of Inpatients With COVID-19: A Computational Psychiatry Approach.

Background: The overload of healthcare systems around the world and the danger of infection have limited the ability of researchers to obtain sufficient and reliable data on psychopathology in hospitalized patients with coronavirus disease 2019 (COVID-19). The relationship between severe acute respiratory syndrome with the coronavirus 2 (SARS-CoV-2) infection and specific mental disturbances remains poorly understood. Aim: To reveal the possibility of identifying the typology and frequency of psychiatric syndromes associated with acute COVID-19 using cluster analysis of discrete psychopathological phenomena. Materials and Methods: Descriptive data on the mental state of 55 inpatients with COVID-19 were obtained by young-career physicians. Classification of observed clinical phenomena was performed with k-means cluster analysis of variables coded from the main psychopathological symptoms. Dispersion analysis with p level 0.05 was used to reveal the clusters differences in demography, parameters of inflammation, and respiration function collected on the basis of the original medical records. Results: Three resulting clusters of patients were identified: (1) persons with anxiety; disorders of fluency and tempo of thinking, mood, attention, and motor-volitional sphere; reduced insight; and pessimistic plans for the future (n = 11); (2) persons without psychopathology (n = 37); and (3) persons with disorientation; disorders of memory, attention, fluency, and tempo of thinking; and reduced insight (n = 7). The development of a certain type of impaired mental state was specifically associated with the following: age, lung lesions according to computed tomography, saturation, respiratory rate, C-reactive protein level, and platelet count. Conclusion: Anxiety and/or mood disturbances with psychomotor retardation as well as symptoms of impaired consciousness, memory, and insight may be considered as neuropsychiatric manifestations of COVID-19 and should be used for clinical risk assessment.