Anton Matviichuk ¹ Viktoriia Yerokhovych ¹ Sergii Zemskov ¹ Yeva Ilkiv ¹ Vitaly Gurianov ¹ Zlatoslava Shaienko ² Tetyana Falalyeyeva ^3,4 Oksana Sulaieva ^1,4 Nazarii Kobyliak ^1,4*

• ¹ Bogomolets National Medical University, Kyiv, Ukraine
• ² Poltava State Medical University, Poltava, Ukraine
• ³ Taras Shevchenko National University of Kyiv, Kyiv, Ukraine
• ⁴ Medical Laboratory CSD, Kyiv, Ukraine

Introduction: Post-COVID-19 syndrome (PCS) is a severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection-associated chronic condition characterized by long-term violations of physical and mental health. People with type 2 diabetes (T2D) are at high risk for severe COVID-19 and PCS. Aim: The current study aimed to define the predictors of PCS development in people with T2D for further planning of preventive measures and improving patient outcomes. Materials and methods: The data were collected through the national survey targeting persons with T2D concerning the history of COVID-19 course and signs and symptoms that developed during or after COVID-19 and continued for more than 12 weeks and were not explained by an alternative diagnosis. In total, 469 patients from different regions of Ukraine were enrolled in the study. Among them, 227 patients reported PCS development (main group), while 242 patients did not claim PCS symptoms (comparison group). Stepwise multivariate logistic regression and probabilistic neural network (PNN) models were used to select independent risk factors. Results: Based on the survey data, 8 independent factors associated with the risk of PCS development in T2D patients were selected: newly diagnosed T2D (OR 4.86; 95% CI 2.55–9.28; p<0.001), female sex (OR 1.29; 95% CI 0.86–1.94; p=0.220), COVID-19 severity (OR 1.35 95% CI 1.05–1.70; p=0.018), myocardial infarction (OR 2.42 95% CI 1.26–4.64; p=0.002) and stroke (OR 3.68 95% CI 1.70–7.96; p=0.001) in anamnesis, HbA1c above 9.2% (OR 2.17 95% CI 1.37–3.43; p=0.001), and the use of insulin analogs (OR 2.28 95% CI 1.31–3.94; p=0.003) vs human insulin (OR 0.67 95% CI 0.39–1.15; p=0.146). Although obesity aggravated COVID-19 severity, it did not impact PCS development. In ROC analysis, the 8-factor multilayer perceptron (MLP) model exhibited better performance (AUC 0.808; 95% CІ 0.770–0.843), allowing the prediction of the risk of PCS development with a sensitivity of 71.4%, specificity of 76%, PPV of 73.6% and NPV of 73.9%. Conclusions: Patients who were newly diagnosed with T2D, had HbA1c above 9.2%, had previous cardiovascular or cerebrovascular events, and had severe COVID-19 associated with mechanical lung ventilation were at high risk for PCS.