EUVIMEDDescobrir EUVIMED
BeobachtungsstudieKohortenstudieOpen access

AI Predictive Model of Mortality and Intensive Care Unit Admission in the COVID-19 Pandemic: Retrospective Population Cohort Study of 12,000 Patients

KI-Vorhersagemodell für Mortalität und Aufnahme auf die Intensivstation in der COVID-19-Pandemie: retrospektive populationsbasierte Kohortenstudie mit 12.000 Patient:innen

Jose Manuel Ruiz Giardin iD, Óscar Garnica iD, Nieves Mesa Plaza iD, Juan Víctor SanMartín López iD, Ana Farfán Sedano iD, Elena Madroñal Cerezo iD, Miguel Ángel Duarte Millán iD, Aida Izquierdo Martínez iD, Luis Rivas iD, Marta Rivilla iD, Alejandro Morales Ortega iD, Begoña Frutos Pérez iD and 34 more

Journal of Medical Internet Research · 2025-07-10 · 27 · e70674

DOI: 10.2196/70674 ↗

Auf einen Blick

Das KI-Modell sagte Mortalität und ICU-Aufnahme bei SARS-CoV-2-Infektion vor der Behandlung gut voraus. Die Leistung blieb über verschiedene Pandemiephasen stabil.

Fragestellung

Lässt sich das individuelle Risiko für Mortalität und ICU-Aufnahme bei SARS-CoV-2-Infektion bereits bei der ersten ärztlichen Einschätzung vor Behandlung vorhersagen?

Methode

Retrospektive populationsbasierte Beobachtungsstudie in Fuenlabrada (Madrid, Spanien) vom 1. Februar 2020 bis 24. Januar 2023; Random Forest, SHAP und Python/scikit-learn; Entwicklung und Validierung mit ED-Daten.

Ergebnisse

11.975 Patient:innen. Mortalitätsmodell: AUC 0,92; Sensitivität 0,89; Spezifität 0,82; PPV 0,35; NPV 0,98. ICU-Modell: AUC 0,89; Sensitivität 0,75; Spezifität 0,88; PPV 0,37; NPV 0,98.

Schlussfolgerung der Autor:innen

Die KI-Modelle zeigten über alle Pandemiephasen stabile Kennwerte und könnten laut Autor:innen in frühen präventiven und prädiktiven klinischen Situationen unterstützen.

Kernaussagen

  • Mortalität: AUC 0,92; ICU-Aufnahme: AUC 0,89
  • Validierung: bei Nicht-Hospitalisierten AUC 0,95 für Mortalität
  • Wichtige Prädiktoren: Alter, Charlson-Wert und Tachypnoe

Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.

At a glance

Beobachtungsstudie in Journal of Medical Internet Research, published online on 2025-07-10. Written by Jose Manuel Ruiz Giardin and 45 co-authors. Institution: Medicina Interna-Infecciosas, Hospital Universitario de Fuenlabrada, Fuenlabrada, Spain. Major MEDLINE topics: COVID-19, Intensive Care Units, Artificial Intelligence, Patient Admission. Cited 7 times so far (Crossref); 30 references. The full text is freely available in PubMed Central.

Auf einen BlickResumoSubjectsDados bibliográficosCiteRelated studies

Abstract (original language)

Background

One of the main challenges with COVID-19 has been that although there are known factors associated with a worse prognosis, clinicians have been unable to predict which patients, with similar risk factors, will die or require intensive care unit (ICU) care.

Objective

This study aimed to develop a personalized artificial intelligence model to predict the patient risk of mortality and ICU admission related to SARS-CoV-2 infection during the initial medical evaluation before any kind of treatment.

Methods

It is a population-based, observational, retrospective study covering from February 1, 2020, to January 24, 2023, with different circulating SARS-CoV-2 viruses, vaccinated status, and reinfections. It includes patients attended by the reference hospital in Fuenlabrada (Madrid, Spain). The models used the random forest technique, Shapley Additive Explanations method, and processing with Python (version 3.10.0; Python Software Foundation) and scikit-learn (version 1.3.0). The models were applied to different epidemic SARS-CoV-2 infection waves. Data were collected from 11,975 patients (4998 hospitalized and 6737 discharged). Predictive models were built with records from 4758 patients and validated with 6977 patients after evaluation in the emergency department. Variables recorded were age, sex, place of birth, clinical data, laboratory results, vaccination status, and radiologic data at admission.

Results

The best mortality predictor achieved an area under the receiver operating characteristic curve (AUC) of 0.92, sensitivity of 0.89, specificity of 0.82, positive predictive value (PPV) of 0.35, and mean negative predictive value (NPV) of 0.98. The ICU admission predictor had an AUC of 0.89, sensitivity of 0.75, specificity of 0.88, PPV of 0.37, and NPV of 0.98. During validation, the mortality model exhibited good performance for the nonhospitalized group, achieving an AUC of 0.95, sensitivity of 0.88, specificity of 0.98, PPV of 0.21, and NPV of 0.99, predicting the death of 30 of 34 patients who were not hospitalized. For the hospitalized patients, the mortality model achieved an AUC of 0.85, sensitivity of 0.86, specificity of 0.74, PPV of 0.24, and NPV of 0.98. The model for predicting ICU admission had an AUC of 0.82, sensitivity of 1.00, specificity of 0.59, PPV of 0.05, and NPV of 1.00. The models’ metrics presented stability along all pandemic waves. Key mortality predictors included age, Charlson value, and tachypnea. The worse prognosis was linked to high values in urea, erythrocyte distribution width, oxygen demand, creatinine, procalcitonin, lactate dehydrogenase, heart failure, D-dimer, oncological and hematological diseases, neutrophil, and heart rate. A better prognosis was linked to higher values of lymphocytes and systolic and diastolic blood pressures. Partial or no vaccination provided less protection than full vaccination.

Conclusions

The artificial intelligence models demonstrated stability across pandemic waves, indicating their potential to assist in personal health services during the 3-year pandemic, particularly in early preventive and predictive clinical situations.

Subjects

MeSH terms (MEDLINE)

COVID-19Intensive Care UnitsArtificial IntelligencePatient AdmissionHumansRetrospective StudiesMaleFemaleMiddle AgedAgedSpainSARS-CoV-2PandemicsPrognosisHospitalizationAdult

Author keywords

COVID-19SARS-CoV-2artificial intelligencedeathintensive care unitmortalitypopulation studypredictive modelrandom forest

Authors and institutions

  • Medicina Interna-Infecciosas, Hospital Universitario de Fuenlabrada, Fuenlabrada, Spain.
  • CIBERINFECT, Centro de Investigación Biomédica en Red, Madrid, Spain.
  • Departamento de Arquitectura de Computadores, Universidad Complutense de Madrid, Madrid, Spain.
  • Medicina Interna, Hospital Universitario de Fuenlabrada, Fuenlabrada, Spain.
  • Urgencias, Hospital Universitario de Fuenlabrada, Fuenlabrada, Spain.
  • Farmacia Hospitalaria, Hospital Universitario de Fuenlabrada, Fuenlabrada, Spain.

ORCID profiles: Jose Manuel Ruiz Giardin, Óscar Garnica, Nieves Mesa Plaza, Juan Víctor SanMartín López, Ana Farfán Sedano, Elena Madroñal Cerezo, Miguel Ángel Duarte Millán, Aida Izquierdo Martínez, Luis Rivas, Marta Rivilla, Alejandro Morales Ortega, Begoña Frutos Pérez, Cristina De Ancos Aracil, Ruth Calderón, Guillermo Soria Fernandez, Jorge Marrero Francés, David Bernal Bello, Jose Ángel Satué Bartolomé, María Toledano Macías, Sara Piedrabuena García, Marta Guerrero Santillán, Rafael Cristóbal, Belen Mora, Laura Velázquez Ríos, Vanesa García de Viedma, Paula Cuenca Ruiz, Ibone Ayala Larrañaga, Lorena Carpintero, Celia Lara, Alvaro Ricardo Llerena, Virginia García Bermúdez, Gema Delgado Cárdenas, Paloma Pardo Rovira, Elena Tejero Sánchez, Maria Jesús Domínguez García, Carolina Mariño, Cristina Bravo, Ana Ontañon, Mario García, Jose Ignacio Hidalgo Pérez, Santiago Prieto Menchero, Natalia González Pereira, Sonia Gonzalo Pascua, Jorge Tarancón Rey, Luis Antonio Lechuga Suárez

Cite

APA

Ruiz Giardin, J. M., Garnica, Ó., Mesa Plaza, N., SanMartín López, J. V., Farfán Sedano, A., Madroñal Cerezo, E., Duarte Millán, M. Á., Izquierdo Martínez, A., Rivas, L., Rivilla, M., Morales Ortega, A., Frutos Pérez, B., De Ancos Aracil, C., Calderón, R., Soria Fernandez, G., Marrero Francés, J., Bernal Bello, D., Satué Bartolomé, J. Á., Toledano Macías, M., …, & FUENCOVID (2025). AI Predictive Model of Mortality and Intensive Care Unit Admission in the COVID-19 Pandemic: Retrospective Population Cohort Study of 12,000 Patients. Journal of Medical Internet Research, 27, e70674. https://doi.org/10.2196/70674

Vancouver

Ruiz Giardin JM, Garnica Ó, Mesa Plaza N, SanMartín López JV, Farfán Sedano A, Madroñal Cerezo E, et al. AI Predictive Model of Mortality and Intensive Care Unit Admission in the COVID-19 Pandemic: Retrospective Population Cohort Study of 12,000 Patients. Journal of Medical Internet Research. 2025;27:e70674. doi:10.2196/70674

BibTeX
@article{RuizGiardin2025,
  author = {Ruiz Giardin, Jose Manuel and Garnica, Óscar and Mesa Plaza, Nieves and SanMartín López, Juan Víctor and Farfán Sedano, Ana and Madroñal Cerezo, Elena and Duarte Millán, Miguel Ángel and Izquierdo Martínez, Aida and Rivas, Luis and Rivilla, Marta and Morales Ortega, Alejandro and Frutos Pérez, Begoña and De Ancos Aracil, Cristina and Calderón, Ruth and Soria Fernandez, Guillermo and Marrero Francés, Jorge and Bernal Bello, David and Satué Bartolomé, Jose Ángel and Toledano Macías, María and Piedrabuena García, Sara and Guerrero Santillán, Marta and Cristóbal, Rafael and Mora, Belen and Velázquez Ríos, Laura and García de Viedma, Vanesa and Cuenca Ruiz, Paula and Ayala Larrañaga, Ibone and Carpintero, Lorena and Lara, Celia and Llerena, Alvaro Ricardo and García Bermúdez, Virginia and Delgado Cárdenas, Gema and Pardo Rovira, Paloma and Tejero Sánchez, Elena and Domínguez García, Maria Jesús and Mariño, Carolina and Bravo, Cristina and Ontañon, Ana and García, Mario and Hidalgo Pérez, Jose Ignacio and Prieto Menchero, Santiago and González Pereira, Natalia and Gonzalo Pascua, Sonia and Tarancón Rey, Jorge and Lechuga Suárez, Luis Antonio and FUENCOVID, FUENCOVID},
  title = {AI Predictive Model of Mortality and Intensive Care Unit Admission in the COVID-19 Pandemic: Retrospective Population Cohort Study of 12,000 Patients},
  journal = {Journal of Medical Internet Research},
  year = {2025},
  volume = {27},
  pages = {e70674},
  doi = {10.2196/70674}
}
Start for free →