Цель: разработать и валидировать модель прогнозирования неблагоприятного исхода у госпитализированных пациентов с COVID-19 и лечением ремдесивиром.
Методы. В исследование включены пациенты в возрасте от 18 лет и старше с диагнозом COVID-19 и лечением ремдесивиром (N = 1428). Набор данных был разделен на выборки обучения моделей, внутренней и внешней валидации. Отбор предикторов неблагоприятного исхода осуществляли методом LASSO-регрессии. Отобранные предикторы включались в модель множественной логистической регрессии. Оптимальные пороги классификации определяли методом Юдена на выборке внутренней валидации. Оценку производительности модели проводили на выборке внешней валидации (n = 487), рассчитывая AUC, чувствительность, специфичность. Целесообразность применения модели в клинической практике оценивали с помощью анализа кривых принятия решений (DCA, decision curves analysis). Статистический анализ выполнен в R 4.4.1 с пакетами tidyr, dplyr, comorbidity, glmnet, caret, boot, calibrationCurves, pROC, rms, dcurves.
Результаты. В качестве наиболее стабильных предикторов отобраны С-реактивный белок, лактатегидрогеназа, Д-димеры. По результатам множественной логистиче- ской регрессии отношения шансов для летального исхода составили: С-реактивный белок (на 10 мг/л) – в 1,097, p < 0,001; ЛДГ (на 50 Ед/л) – 1,276, p < 0,001; Д-димеры > 400 нг/мл – 5,829, p < 0,001. Оптимальная точка классификации на выборке внутренней валидации составила 4,9 %. На внешних данных AUC модели составила 89,5 %, чувствительность модели – 93,8 %, а специфичность – 72,9 %. Применение модели характеризуется клиническим преимуществом при любых порогах классификации по сравнению с альтернативными стратегиями наблюдения пациентов.
Заключение. Разработан и валидирован метод прогнозирования неблагоприятного исхода COVID-19 у пациентов с лечением ремдесивиром, который представлен в виде номограммы.
Background. Remdesivir is used for antiviral treatment in patients with COVID-19. In this population, early quantitative assessment of mortality risk remains a clinical priority. Aim of the study is to develop and validate prognostic model for predicting risk of lethal outcomes in hospitalized COVID-19 patients treated with remdesivir.
Methods. The study included patients aged 18 years and older with a diagnosis of COVID-19 and treatment with remdesivir (N = 1428). The observations were divided into training, internal validation and external validation sets. Following selection using LASSO regression, predictors were included in multivariable logistic regression model. Optimal classification thresholds were determined with Youden’s metric on internal validation set. Model performance was assessed in the external validation cohort (n = 487) using the AUC, sensitivity, and specificity. Clinical utility of the model was assessed using decision curves analysis. Statistical analysis was performed in R version 4.4.1 with libraries tidyr, dplyr, comorbidity, glmnet, caret, boot, calibrationCurves, pROC, rms, dcurves.
Results. LASSO regression selected C-reactive protein, lactate dehydrogenase, and D-dimer
as most stable predictors. According to the multivariable logistic regression model, C-reactive
protein (per 10 mg/L), OR 1.098, p < 0.001; lactate dehydrogenase (per 50 U/L), OR = 1.276,
p < 0.001; and D-dimers > 400 ng/mL, OR = 5.829, p < 0.001, were independently associated
with lethal outcome. Optimal classification threshold on internal validation set was 4.9 %.
On external validation cohort, the model achieved AUC of 89.5 %, with sensitivity of 93.8 %
and specificity of 72.9 %. Decision curve analysis demonstrated that the model provides a clinical
net benefit across all threshold probabilities compared to alternative management strategies.
Conclusion. We developed and validated prognostic model for predicting lethal outcomes
in COVID-19 patients receiving remdesivir treatment. The model is presented as a nomogram
for convenient clinical application.
- WHO COVID-19 dashboard / World Health Organization. – Electronic data. – Geneva: WHO, 2026. – Available at: https://data.who.int/dashboards/covid19/cases. – Accessed: 09.03.2026.
- COVID-19 epidemiological update – 14 March 2025 / World Health Organization. – Geneva: WHO, 2025. – 31 p. – Available at: https://www.who.int/publications/m/item/covid19epidemiologicalupdateedition177. – Accessed: 09.03.2026.
- Remdesivir for the Treatment of Covid19 – Final Report / J. H. Beigel, K. M. Tomashek, L. E. Dodd [et al.] // The New England Journal of Medicine. – 2020. – Vol. 383, № 19. – P. 1813–1826. – doi: 10.1056/NEJMoa2007764.
- Infectious Diseases Society of America Guidelines on the Treatment and Management of Patients With COVID19 (September 2022) / A. Bhimraj, R. L. Morgan, A. H. Shumaker [et al.] // Clinical Infectious Diseases. – 2024. – Vol. 78, № 7. – P. e250–e349. – doI: 10.1093/cid/ciac724.
- Рекомендации (временные) об организации оказания медицинской помощи пациентам с инфекцией COVID19: [утв. Приказом Министерства здравоохранения Респ. Беларусь, 21 июля 2021 г., № 900]. – Минск, 2021. – 75 с. Rekomendacii (vremennye) ob organizacii okazaniya medicinskoj pomoshchi pacientam s infekciej COVID19 [Recommendations (temporary) on the organization of medical care for patients with COVID19 infection]: [utv. Prikazom Ministerstva zdravoohraneniya Resp. Belarus', 21 iyulya 2021 g., № 900]. – Minsk, 2021. – 75 s. [In Russian].
- Remdesivir for 5 or 10 Days in Patients with Severe Covid19 / J. D. Goldman, D. C. B. Lye, D. S. Hui [et al.] // The New England Journal of Medicine. – 2020. – Vol. 383, № 19. – P. 1827–1837. – doi: 10.1056/NEJMoa2015301.
- Bálint, G. Omicron: increased transmissibility and decreased pathogenicity / G. Bálint, B. VörösHorváth, A. Széchenyi // Signal Transduction and Targeted Therapy. – 2022. – Vol. 7, № 1. – P. 151. – doi: 10.1038/s41392022010098.
- Impact of SARSCoV2 Omicron and Delta variants in patients requiring intensive care unit (ICU) admission for COVID19, Northern Italy, December 2021 to January 2022 / A. Piralla, F. Mojoli, L. Pellegrinelli [et al.] // Respiratory Medicine and Research. – 2023. – Vol. 83. – P. 100990. – doi: 10.1016/j.resmer. 2023.100990.
- Prediction of inhospital mortality with machine learning for COVID19 patients treated with steroid and remdesivir / T. Kuno, Y. Sahashi, S. Kawahito [et al.] // Journal of Medical Virology. – 2022. – Vol. 94, № 3. – P. 958–964. – doi: 10.1002/jmv.27393.
- A Systematic Review of Predictor Composition, Outcomes, Risk of Bias, and Validation of COVID19 Prognostic Scores / K. S. Appel, R. Geisler, D. Maier [et al.] // Clinical Infectious Diseases. – 2024. – Vol. 78, № 4. – P. 889–899. – doi: 10.1093/ cid/ciad618.
- Biomarkers and shortterm prognosis in COVID19 / Ó. M. Peiró, A. Carrasquer, R. SánchezGimenez [et al.] // Biomarkers. – 2021. – Vol. 26, № 2. – P. 119–126. – doi: 10.1080/1354750X.2021.1874052.
- Prognostic implications of neutrophillymphocyte ratio in COVID19 / S. Jimeno, P. S. Ventura, J. M. Castellano [et al.] // European Journal of Clinical Investigation. – 2021. – Vol. 51, № 1. – P. e13404. – doi: 10.1111/eci.13404.
- Abnormal Liver Tests in COVID19: A Retrospective Observational Cohort Study of 1,827 Patients in a Major U. S. Hospital Network / M. A. Hundt, Y. Deng, M. M. Ciarleglio [et al.] // Hepatology. – 2020. – Vol. 72, № 4. – P. 1169–1176. – doi: 10.1002/hep.31487.
- Association between thrombocytopenia and 180day prognosis of COVID19 patients in intensive care units: A twocenter observational study / Y. Zhu, J. Zhang, Y. Li [et al.] // PLoS One. – 2021. – Vol. 16, № 3. – P. e0248671. – doi: 10.1371/journal.pone.0248671.
- Anemia is associated with severe illness in COVID19: A retrospective cohort study / Z. Tao, J. Xu, W. Chen [et al.] // Journal of Medical Virology. – 2021. – Vol. 93, № 3. – P. 1478–1488. – doi: 10.1002/jmv.26444.
- Preexisting Comorbidities Predicting COVID19 and Mortality in the UK Biobank Community Cohort / J. L. Atkins, J. A. H. Masoli, J. Delgado [et al.] // The Journals of Gerontology. Series A:Biological Sciences and Medical Sciences. – 2020. – Vol. 75, № 11. – P. 2224–2230. – doi: 10.1093/gerona/glaa183.
- Krasko, O. V., Reutovich M. Yu., Ivanov A. V. Prediction and decisionmaking based on nonlinear risks model in stomach cancer treatment // Informatics. – 2024. – Т. 21, № 1. – С. 65–82. – doi: 10.37661/1816030120242116582.
- Mehta, P. Is severe COVID19 a cytokine storm syndrome: a hyperinflammatory debate / P. Mehta, D. C. Fajgenbaum // Current Opinion in Rheumatology. – 2021. – Vol. 33, № 5. – P. 419–430. – doi: 10.1097/BOR.0000000000000822.
- Lactate dehydrogenase, an independent risk factor of severe COVID19 patients: a retrospective and observational study / Y. Han, H. Zhang, S. Mu [et al.] // Aging (Albany NY). – 2020. – Vol. 12, № 12. – P. 11245–11258. – doi: 10.18632/aging.103372.