Respiratory failure ae cause diagnostic respiratory failure markers, models and algorithms for predicting the course of COVID-19

Introduction

Artificial intelligence has been widely used in the fight against COVID-19. Very few studies have focused on creating algorithms and IT products to predict the length of hospital stay of COVID-19 patients, primarily using functional markers of respiratory failure.

Aim: this study aims to identify the most representative functional markers of respiratory failure to create algorithms for predicting the length of hospital stay (LOS) of patients with COVID-19.

Materials and methods

A comparative analysis was performed using a database of 487 patients with COVID-19 hospitalized from December 2020 to May 2022, in whom a surrogate oxygenation index (SpO₂/FiO₂), ROX ((SpO₂/FiO₂)/respiratory rate) were measured during the first three days of treatment, and spirometry (n = 103, 62 ± 13 years) was also performed simultaneously to calculate “ventilatory reserve” (VR = MVV/MV) or measurement (n = 384, 61 ± 16 years) of maximal inspiratory pressure (MIP) and maximal expiratory pressure (MEP). Eleven logistic regression (LR) models were constructed/ selected, followed by ROC analysis and description of sensitivity (SE), specificity (SP), area under the operating characteristic curve (AUC) and accuracy (Ac). The least squares method and the coefficient of determination R2 was used to assess the possibility of constructing algorithms for calculating the prognosis of the Prolonged Length of Stay (PLoS) with the selected markers.

Results

The constructed algorithms for calculating the probability of discharge within seven days using ROX, SpO2/FiO2, and ventilation reserve showed that a fifth-degree polynomial approximating function discriminates fairly well against patients at risk of prolonged hospitalization (R2 = 0.87, 0.82, and 0.85, respectively).

Conclusion: the most effective functional markers for creating predictive algorithms for the duration of hospitalization of patients with COVID-19 are the integral SpO2/FiO2, ROX indicators, combined with the indicators of reserve and strength of the VR and MIP respiratory muscles, the use of which is promising in the development of tools for predicting respiratory failure of various etiologies using artificial intelligence.

1. Piccialli, F. The Role of Artificial Intelligence in Fighting the COVID-19 Pandemic. / F. Piccialli, V. S di Cola, F. Giampaolo, S. Cuomo // Inf Syst Front. 2021;23(6):1467–1497. doi:10.1007/s10796-021-10131-x

2. Negrini, D. Artificial intelligence at the time of COVID-19: who does the lion’s share? / D. Negrini, E. Danese, B. M. Henry, G. Lippi, M. Montagnana// Clin Chem Lab Med. 2022;60(12):1881–1886. Published 2022 Apr 25. doi:10.1515/cclm-2022-0306

3. Kas’yanenko, K. V. Prognoz stepeni tyazhesti techeniya SARS-CoV-2-infekcii u lic molodogo vozrasta s primeneniem metodov iskusstvennogo intellekta. / Kas’yanenko K. V., Kozlov K. V., ZHdanov K. V., Lapikov I. I., Belikov V. V. // ZHurnal infektologii. 2022;14(5). https://doi.org/10.22625/2072-6732-2022-14-5-14-25

4. Karonova, T. L. Ispol’zovanie iskusstvennogo intellekta u bol’nyh s novoj koronavirusnoj infekciej dlya prognozirovaniya techeniya zabolevaniya v usloviyah infekcionnogo stacionara / T. L. Karonova, I. N. Korsakov, A. A. Mihajlova, D. I. Lagutina, A. T. CHernikova, M. A. Vashukova, M. V. Smol’nikova, D. A. Gusev, A. O. Konradi, E. V. SHlyahto // ZHurnal infektologii. 2023;15(3):60–66. https://doi.org/10.22625/2072-6732-2023-15-3-60-66

5. Afrash, M. R. Predictive Modeling of Hospital Length of Stay in COVID-19 Patients Using Machine Learning Algorithms / M. R. Afrash, H. Kazemi-Arpanahi, P. Ranjbar, R. Noupor, M. Saki, M. Amraei, M. Shanbehzadeh // Journal of Medicinal and Chemical Sciences. 2021., 4(5), 525–537. doi: 10.26655/JMCHEMSCI.2021.5.15

6. Archer, S. L. Differentiating COVID-19 Pneumonia From Acute Respiratory Distress Syndrome and High Altitude Pulmonary Edema: Therapeutic Implications.Circulation / S. L. Archer, W. W. Sharp, E. K. Weir //Circulationaha. 2020;142(2):101-104. doi:10.1161.120.047915.

7. Gibson, P. G. COVID-19 acute respiratory distress syndrome (ARDS): Clinical features and differences from typical pre-COVID-19 ARDS / P. G. Gibson, L. Qin, S. H. Puah // Med. J. Aust. 2020;213:54– 56.e1. doi: 10.5694/mja2.50674

8. Rahman, A. Silent hypoxia in COVID-19: pathomechanism and possible management strategy / A. Rahman, T. Tabassum, Y. Araf, A. Al Nahid, M. A. Ullah, M. J. Hosen. / Mol Biol Rep. 2021;48(4):3863–3869. doi:10.1007/s11033-021-06358-1

9. Pan’ko, E. S. Prediktory dlitel’nosti stacionarnogo lecheniya ostroj fazy infekcionnogo processa, vyzvannogo COVID-19. / E. S. Pan’ko, S. V. Zhavoronok, A. M. Solovchuk, S. C. Pan’ko, S. V. Pan’ko». ZHurnal infektologii. 2023;15(1):86-92. https://doi.org/10.22625/2072-6732-2023-15-1-86-92

10. Pan’ko, E. S. Pokazateli sily dyhatel’nyh myshc i oksigenacii dlya prognozirovaniya dlitel’nosti gospitalizacii pacientov s COVID-19. / E. S. Pan’ko, S. V. ZHavoronok, A. M. Solovchuk, S. V. Pan’ko. / ZHurnal infektologii. 2024;16(4):71-79. https://doi.org/10.22625/2072-6732-2024-16-3-71-79

11. Wang, L. Artificial Intelligence for COVID-19: A Systematic Review. / L. Wang, Y. Zhang, D. Wang, X. Tong, T. Liu, S. Zhang, J. Huang, L. Zhang, L. Chen, H. Fan, M. Clarke / Front Med (Lausanne). 2021 Sep 30;8: 704256. doi: 10.3389/fmed.2021.704256. PMID: 34660623; PMCID: PMC8514781

12. Daghistani, T. A. Predictors of in-hospital length of stay among cardiac patients: A machine learning approach. T. A. Daghistani / Int J Cardiol. 2019; 288:140–147. doi:10.1016/j.ijcard.2019.01.046.

13. Muhlestein, W. E. Predicting Inpatient Length of Stay After Brain Tumor Surgery: Developing Machine Learning Ensembles to Improve Predictive Performance / W. E. Muhlestein, D. S. Akagi, J. M. Davies, L. B. Chambless / Neurosurgery. 2019;85(3):384–393. doi:10.1093/neuros/nyy343

14. Staziaki, P. V. Machine learning combining CT findings and clinical parameters improves prediction of length of stay and ICU admission in torso trauma. / P. V. Staziaki, D. Wu, J. C. Rayan, et al. / Eur Radiol. 2021;31(7):5434–5441. doi:10.1007/s00330-020-07534-w

15. Pei, J. Prediction of Patients’ Length of Stay at Hospital During COVID-19 Pandemic / J. Pei, X. Lin, Q. Chen / J. Phys.: Conf. Ser., 2021. 1802 032038 DOI: 10.1088/1742-6596/1802/3/032038

16. Liang, W. China Medical Treatment Expert Group for COVID-19. Development and Validation of a Clinical Risk Score to Predict the Occurrence of Critical Illness in Hospitalized Patients With COVID-19. / W. Liang, H. Liang, L. Ou, B. Chen, A. Chen, W. Guan, L. Sang, J. Lu, Y. Xu, G. Chen, H. Guo, J. Guo, Z. Chen, Y. Zhao, S. Li, N. Zhang, N. Zhong, J. He. / Intern Med. 2020 Aug 1;180(8):1081-1089. doi: 10.1001/jamainternmed.2020.2033. PMID: 32396163; PMCID: PMC7218676.

17. Dan, T. Machine Learning to Predict ICU Admission, ICU Mortality and Survivors’ Length of Stay among COVID-19 Patients: Toward Optimal Allocation of ICU Resources. / T. Dan et al / IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Seoul, Korea (South), 2020, pp. 555–561, doi: 10.1109/BIBM49941.2020.9313292.

18. Rehm, G. B. Leveraging IoTs and Machine Learning for Patient Diagnosis and Ventilation Management in the Intensive Care Unit / G. B. Rehm, S. H. Woo, X. L. Chen, et al. / IEEE Pervasive Comput. 2020;19(3):68–78. doi:10.1109/mprv.2020.2986767

19. Amiri, P. Prediction of mortality risk and duration of hospitalization of COVID-19 patients with chronic comorbidities based on machine learning algorithms / P. Amiri, M. Montazeri, F. Ghasemian, et al. // Digital health. 2023;9. doi:10.1177/20552076231170493

20. Ren, S. Machine learning based algorithms to impute PaO2 from SpO2 values and development of an online calculator / S. Ren, J. A. Zupetic, M. Tabary, et al. / Sci Rep. 2022;12(1):8235. Published 2022 May 17. doi:10.1038/s41598-022-12419