Development of a prediction model for 30-day COVID-19 hospitalization and death in a national cohort of Veterans Health Administration patients – March 2022 - April 2023 DOI Open Access
David Bui, Kristina L. Bajema, Yuan Huang

et al.

medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Nov. 18, 2023

ABSTRACT Objective Develop models to predict 30-day COVID-19 hospitalization and death in the Omicron era for clinical research applications. Material Methods We used comprehensive electronic health records from a national cohort of patients Veterans Health Administration (VHA) who tested positive SARS-CoV-2 between March 1, 2022, 31, 2023. Full incorporated 84 predictors, including demographics, comorbidities, receipt vaccinations anti-SARS-CoV-2 treatments. Parsimonious included 19 predictors. created or death, hospitalization, all-cause mortality. Super Learner ensemble machine learning algorithm fit prediction models. Model performance was assessed with area under receiver operating characteristic curve (AUC), Brier scores, calibration intercepts slopes 20% holdout dataset. Results Models were trained on 198,174 patients, whom 8% hospitalized died within 30 days testing positive. AUCs full ranged 0.80 (hospitalization) 0.91 (death). scores close 0, lowest error mortality model (Brier score: 0.01). All three well calibrated <0.23 <1.05. performed comparably Discussion These may be risk stratification inform treatment identify high-risk inclusion trials. Conclusions developed that accurately estimate following emergence variant setting antiviral

Language: Английский

External Validation of the 4C (Coronavirus Clinical Characterization Consortium) Mortality Score in a Teaching Hospital in Brazil DOI Open Access
Katelyn A. Bruno,

Henrique Thadeu Periard Mussi,

Alessandro Bruno

et al.

Cureus, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 2, 2025

Background The 4C (Coronavirus Clinical Characterization Consortium) Mortality Score has demonstrated good discrimination in COVID-19 but not been widely validated Brazil. is a clinical tool developed during the pandemic to predict in-hospital mortality for patients admitted with COVID-19. It was derived from large dataset of hospitalized United Kingdom and provides simple yet effective way stratify based on their risk death. Objective This study aimed determine accuracy university teaching hospital. Methods observational, longitudinal, retrospective, conducted 180-bed hospital Rio de Janeiro, We included all followed them until discharge. calculated age, sex, Charlson index, respiratory rate, peripheral oxygen saturation (room air), Glasgow Coma Scale, serum urea, C-reactive protein (CRP) level. primary outcome mortality. Results 208 participants, median age 63 years. Among them, 111 (53%) were male; 52 (25%) had cardiovascular disease, 83 (39%) cancer. 39.9%. Independent predictors hemoglobin, CRP, mechanical ventilation, need vasopressors. Score's area under receiver operating characteristic curve (AUC-ROC) 89.9%. Conclusion excellent population.

Language: Английский

Citations

0
Predicting the risk of intensive care unit admission in patients with COVID-19 presenting in the emergency room: Development and evaluation of the CROSS score DOI Creative Commons

Weiwei Xiang,

Fridolin Steinbeis, Kiret Dhindsa

et al.

Clinical Infectious Diseases, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 10, 2025

Abstract Background Existing risk evaluation tools underperform in predicting intensive care unit (ICU) admission for patients with the Coronavirus Disease 2019 (COVID-19). This study aimed to develop and evaluate an accurate calculator-free clinical tool ICU at emergency room (ER) presentation. Methods Data from COVID-19 a nationwide German cohort (March 2020-January 2023) were analyzed. Candidate predictors selected based on literature expertise. A score, within seven days of ER presentation, was developed using elastic net logistic regression northern (derivation cohort), evaluated southern (evaluation cohort) externally validated Colombian cohort. Performance through discrimination, calibration, utility against existing tools. Results rates 30.8% cohort, n=1295, median age 60, 38.1% female), 28.1% n=1123, 58, 36.9% 30.3% (Colombian n=780, 57, 38.8% female). The 11-point CROSS Confusion, Respiratory rate, Oxygen Saturation (with or without concurrent supplemental oxygen), oxygen Supplementation, demonstrated good discrimination (area under curve (AUC): 0.77 cohort; 0.69 superior compared Mortality-predicting performed poorly COVID-19. Conclusions score effectively predicts ER. Further studies are needed assess its generalizability other settings. not recommended prediction.

Language: Английский

Citations

0
Leveraging near-real-time patient and population data to incorporate fluctuating risk of severe COVID-19: development and prospective validation of a personalised risk prediction tool DOI Creative Commons
Kaitlin N. Swinnerton, Nathanael R. Fillmore,

Austin D Vo

et al.

EClinicalMedicine, Journal Year: 2025, Volume and Issue: 81, P. 103114 - 103114

Published: Feb. 21, 2025

Language: Английский

Citations

0
Impact of Surge Strain and Pandemic Progression on Prognostication by an Established COVID-19–Specific Severity Score DOI Creative Commons
Christina Yek, Jing Wang, Jonathan Fintzi

et al.

Critical Care Explorations, Journal Year: 2023, Volume and Issue: 5(12), P. e1021 - e1021

Published: Dec. 1, 2023

Many U.S. State crisis standards of care (CSC) guidelines incorporated Sequential Organ Failure Assessment (SOFA), a sepsis-related severity score, in pandemic triage algorithms. However, SOFA performed poorly COVID-19. Although disease-specific scores may perform better, their prognostic utility over time and overcrowded settings remains unclear.We evaluated prognostication by the modified 4C (m4C) COVID-19-specific prognosticator that demonstrated good predictive capacity early pandemic, as potential tool to standardize across hospital-surge environments.Retrospective observational cohort study.Two hundred eighty-one hospitals an administrative healthcare dataset.A total 298,379 hospitalized adults with COVID-19 were identified from March 1, 2020, January 31, 2022. m4C calculated admission diagnosis codes, vital signs, laboratory values.Hospital-surge index, severity-weighted measure caseload, was for each hospital-month. Discrimination in-hospital mortality surge index-adjusted models measured area under receiver operating characteristic curves (AUC). Calibration assessed training on waves measuring fit (deviation bisector) subsequent waves.From 2020 2022, admitted 281 hospitals. adequately discriminated wave 1 (AUC 0.779 [95% CI, 0.769-0.789]); discrimination lower (wave 2: 0.772 0.765-0.779]; 3: 0.746 0.743-0.750]; delta: 0.707 0.702-0.712]; omicron: 0.729 0.721-0.738]). reduced calibration contemporaneous persisted despite periodic recalibration. Performance characteristics similar without adjustment surge.Mortality prediction score remained robust strain, making it attractive when is most needed. performance has deteriorated recent waves. CSC relying defined prognosticators, especially dynamic disease processes like COVID-19, warrant frequent reappraisal ensure appropriate resource allocation.

Language: Английский

Citations

3
Persistent disabilities 28 months after COVID-19 hospitalization, a prospective cohort study DOI Creative Commons

Bertrand Renaud,

Richard Chocron,

Guillaume Reverdito

et al.

ERJ Open Research, Journal Year: 2024, Volume and Issue: unknown, P. 00104 - 2024

Published: May 16, 2024

Background Limited data are available on long-term respiratory disabilities in patients following acute COVID-19. Patients and Methods This prospective, monocentric, observational cohort study included admitted to our hospital with COVID-19 between March 3 April 24, 2020. Clinical, functional, radiological were collected up 28 months after discharge. Results Among 715 hospitalized for COVID-19, 493 (69.0%) discharged alive. We could access complete medical records 268/493 (54.4%); 138/268 (51.5%) exhibited persistent symptoms agreed the collection follow-up. predominantly male (64.5%), a mean (± sd ) age of 58.9±15.3 years. At last follow-up, leading asthenia (31.5%), dyspnoea (29.8%), neuropsychological (17.7%). Lung function improved visit. Mean diffusing capacity lung carbon monoxide (DLCO) was 77.8% predicted value, total (TLC) 83.5%, O 2 desaturation during exercise (O desaturation) −2.3%. While DLCO over entire period, TLC early phase late phase. Except those comorbidities, only one patient presented minor functional chest alterations at 28-months. Conclusion alive showed clinical symptoms, parameters signs post infection. Persistent consisted mainly dyspnoea, returning normal. One without prior issues moderate pulmonary fibrosis.

Language: Английский

Citations

0
Optimizing Predictive Models in Healthcare Using Artificial Intelligence: A Comprehensive Approach with a COVID-19 Case Study DOI
Juan Carlos Astudillo Sarmiento, Kevin Chamorro, Santiago Ballaz

et al.

Communications in computer and information science, Journal Year: 2024, Volume and Issue: unknown, P. 178 - 192

Published: Oct. 10, 2024

Language: Английский

Citations

0
Impact on clinical guideline adherence of Orient-COVID, a clinical decision support system based on dynamic decision trees for COVID19 management: a randomized simulation trial with medical trainees DOI Creative Commons

M. Jammal,

Antoine Saab, Cynthia Abi Khalil

et al.

International Journal of Medical Informatics, Journal Year: 2024, Volume and Issue: 195, P. 105772 - 105772

Published: Dec. 20, 2024

Citations

0
Development of a prediction model for 30-day COVID-19 hospitalization and death in a national cohort of Veterans Health Administration patients – March 2022 - April 2023 DOI Open Access
David Bui, Kristina L. Bajema, Yuan Huang

et al.

medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Nov. 18, 2023

ABSTRACT Objective Develop models to predict 30-day COVID-19 hospitalization and death in the Omicron era for clinical research applications. Material Methods We used comprehensive electronic health records from a national cohort of patients Veterans Health Administration (VHA) who tested positive SARS-CoV-2 between March 1, 2022, 31, 2023. Full incorporated 84 predictors, including demographics, comorbidities, receipt vaccinations anti-SARS-CoV-2 treatments. Parsimonious included 19 predictors. created or death, hospitalization, all-cause mortality. Super Learner ensemble machine learning algorithm fit prediction models. Model performance was assessed with area under receiver operating characteristic curve (AUC), Brier scores, calibration intercepts slopes 20% holdout dataset. Results Models were trained on 198,174 patients, whom 8% hospitalized died within 30 days testing positive. AUCs full ranged 0.80 (hospitalization) 0.91 (death). scores close 0, lowest error mortality model (Brier score: 0.01). All three well calibrated <0.23 <1.05. performed comparably Discussion These may be risk stratification inform treatment identify high-risk inclusion trials. Conclusions developed that accurately estimate following emergence variant setting antiviral

Language: Английский

Citations

0