The lost horizon and the urgent need to define time in COVID-19 predictive models
Keywords:
algorithms, biomarkers, COVID-19, forecasting, logistic models, predictive value of tests, prognosis, sensitivity and specificity, time factorsAbstract
Predictive models of hyperinflammation in COVID-19 suffer from a critical flaw: the omission of the temporal dimension or prognostic horizon. Although they exhibit high statistical accuracy, this lacks clinical utility if the specific timeframe for which the event is predicted is not defined. The pathophysiology of the disease is dynamic, and the cytokine storm occurs within a specific time window. Predicting hyperinflammation without a defined timeframe (e.g., 48 hours vs. 7 days) prevents the guidance of timely therapeutic interventions, whose effectiveness is maximized when applied early. The predominant methodology, such as logistic regression, oversimplifies this reality. It is proposed to adopt techniques such as survival analysis, which explicitly models the time to the event. Integrating the "when" is crucial to transforming statistical tools into clinically relevant instruments that improve decision-making.
Downloads
References
1. Steyerberg EW. Clinical prediction models: A practical approach to development, validation, and updating. 2a ed. Cham, Suiza: Springer Nature; 2020.
2. Silva MJA, Ribeiro LR, Gouveia MIM, Marcelino BDR, Santos CSD, Lima KVB, Lima LNGC. Hyperinflammatory Response in COVID-19: A Systematic Review [Internet]. Viruses. 2023;15(2):553. DOI: https://doi:10.3390/v15020553
3. Ferrer Castro JE, del Río Caballero G, Amaro Guerra I, Benítez Sánchez E, Rodríguez González Z. Modelo predictivo de hiperinflamación en pacientes con la COVID-19 [Internet]. Rev. cuba. med. Mil. 2025 [acceso: 10/10/20225];54(4):e025076744. Disponible en: https://revmedmilitar.sld.cu/index.php/mil/article/view/76744
4. Wolff RF, Moons KGM, Riley RD, Whiting PF, Westwood M, Collins GS, et al. PROBAST: A Tool to Assess the Risk of Bias and Applicability of Prediction Model Studies [Internet]. Ann Intern Med. 2019;170(1):51-58. DOI: https://doi:10.7326/M18-1376
5. Song Q, Fei W. Evaluation of Sepsis-1 and Sepsis-3 Diagnostic Criteria in Patients with Sepsis in Intensive Care Unit [Internet]. J Healthc Eng. 2023 ;2023:3794886. DOI: https://doi:10.1155/2023/3794886
6. Van Calster B, McLernon DJ, van Smeden M. Calibration: the Achilles heel of predictive analytics [Internet]. BMC Med. 2019; 17:230. DOI: https://doi.org/10.1186/s12916-019-1466-7
7. Kent P, Cancelliere C, Boyle E. A conceptual framework for prognostic research [Internet]. BMC Med Res Methodol. 2020; 20:172. DOI: https://doi.org/10.1186/s12874-020-01050-7
8. Rajasekaran S, Kruse K, Kovey K, Davis AT, Hassan NE, Ndika AN, et al. Therapeutic role of anakinra, an interleukin-1 receptor antagonist, in the management of secondary hemophagocytic lymphohistiocytosis/sepsis/multiple organ dysfunction/macrophage activating syndrome in critically ill children [Internet]. Pediatr Crit Care Med. 2014;15(5):401-8. DOI: https://doi:10.1097/PCC.0000000000000078
9. Wynants L, Van Calster B, Collins GS, Riley RD, Heinze G, Schuit E, et al. Prediction models for diagnosis and prognosis of covid-19: systematic review and critical appraisal [Internet]. BMJ. 2020;369:1328. DOI: https://doi:10.1136/bmj.m1328
10. Laing AG, Lorenc A, Del Molino Del Barrio I, Das A, Fish M, et al. A dynamic COVID-19 immune signature includes associations with poor prognosis [Internet]. Nat Med. 2020;26(10):1623-35. DOI: https://doi:10.1038/s41591-020-1038-6
11. Ouko RK, Mukaka M, Ohuma EO. Joint modelling of longitudinal data: a scoping review of methodology and applications for non-time to event data [Internet]. BMC Med Res Methodol. 2025; 25(1):40. DOI: https://doi:10.1186/s12874-025-02485-6
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Lodixi Cobas Planchez, Natascha Mezquia de Pedro
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
Authors who have publications with this Journal accept the following terms:
- The authors will retain their copyright and guarantee the Journal the right of first publication of their work, which will simultaneously be subject to the Creative Commons Attribution License. The content presented here can be shared, copied and redistributed in any medium or format; Can be adapted, remixed, transformed or created from the material, using the following terms: Attribution (giving appropriate credit to the work, providing a link to the license, and indicating if changes have been made); non-commercial (you cannot use the material for commercial purposes) and share-alike (if you remix, transform or create new material from this work, you can distribute your contribution as long as you use the same license as the original work).
- The authors may adopt other non-exclusive license agreements for the distribution of the published version of the work (for example: depositing it in an institutional electronic archive or publishing it in a monographic volume) as long as the initial publication in this Journal is indicated.
- Authors are allowed and recommended to disseminate their work through the Internet (e.g., in institutional electronic archives or on their website) before and during the submission process, which can produce interesting exchanges and increase citations. of the published work.
