1
|
Lana FCB, Marinho CC, de Paiva BBM, Valle LR, do Nascimento GF, da Rocha LCD, Carneiro M, Batista JDL, Anschau F, Paraiso PG, Bartolazzi F, Cimini CCR, Schwarzbold AV, Rios DRA, Gonçalves MA, Marcolino MS. Unraveling relevant cross-waves pattern drifts in patient-hospital risk factors among hospitalized COVID-19 patients using explainable machine learning methods. BMC Infect Dis 2025; 25:537. [PMID: 40234758 PMCID: PMC12001466 DOI: 10.1186/s12879-025-10766-0] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/30/2024] [Accepted: 03/07/2025] [Indexed: 04/17/2025] Open
Abstract
BACKGROUND Several studies explored factors related to adverse clinical outcomes among COVID-19 patients but lacked analysis of the impact of the temporal data shifts on the strength of association between different predictors and adverse outcomes. This study aims to evaluate factors related to patients and hospitals in the prediction of in-hospital mortality, need for invasive mechanical ventilation (IMV), and intensive care unit (ICU) transfer throughout the pandemic waves. METHODS This multicenter retrospective cohort included COVID-19 patients from 39 hospitals, from March/2020 to August/2022. The pandemic was divided into waves: 10/03/2020-14/11/2020 (first), 15/11/2020-25/12/2021 (second), 26/12/2021-03/08/2022 (third). Patient-related factors included clinical, demographic, and laboratory data, while hospital-related factors covered funding sources, accreditation, academic status, and socioeconomic characteristics. Shapley additive explanation (SHAP) values derived from the predictions of a light gradient-boosting machine (LightGBM) model were used to assess potential risk factors for death, IMV and ICU. RESULTS Overall, 16,958 adult patients were included (median age 59 years, 54.7% men). LightGBM achieved competitive effectiveness metrics across all periods. Temporal drifts were observed due to a decrease in various metrics, such as the recall for the positive class [ICU: 0.4211 (wave 1) to 0.1951 (wave 3); IMV: 0.2089 (wave 1) to 0.0438 (wave 3); death: 0.2711 (wave 1) to 0.1175 (wave 3)]. Peripheral arterial oxygen saturation to the fraction of inspired oxygen ratio (SatO2/FiO2) at admission had great predictive capacity for all outcomes, with an optimal cut-off value for death prediction of 227.78. Lymphopenia had its association strength increased over time for all outcomes, optimal threshold for death prediction of 643 × 109/L. Thrombocytopenia was the most important feature in wave 2 (ICU); overall, values below 143,000 × 109/L were more related to death. CONCLUSION Data drifts were observed in all scenarios, affecting potential predictive capabilities of explainable machine learning methods. Upon admission, SatO2/FiO2 values, platelet and lymphocyte count were significant predictors of adverse outcomes in COVID-19 patients. Overall, inflammatory response markers were more important than clinical characteristics. Limitations included sample representativeness and confounding factors. Integrating the drift's knowledge into models to improve effectiveness is a challenge, requiring continuous updates and monitoring of performance in real-world applications. CLINICAL TRIAL NUMBER Not applicable.
Collapse
Affiliation(s)
| | - Carolina Coimbra Marinho
- Department of Internal Medicine, Medical School & Telehealth Center, University Hospital, Universidade Federal de Minas Gerais, Av. Professor Alfredo Balena, Belo Horizonte, 110, Brazil
| | - Bruno Barbosa Miranda de Paiva
- Computer Science Department, Universidade Federal de Minas Gerais, Av. Presidente Antônio Carlos, Belo Horizonte, 6627, Brazil
| | - Lucas Rocha Valle
- Universidade Federal de Minas Gerais, Av. Presidente Antônio Carlos, Belo Horizonte, 6627, Brazil
| | | | | | - Marcelo Carneiro
- Hospital Santa Cruz. R. Fernando Abott, Santa Cruz do Sul, 174, Brazil
| | | | - Fernando Anschau
- Hospital Nossa Senhora da Conceição, Av. Francisco Trein, Porto Alegre, 326, Brazil
| | | | - Frederico Bartolazzi
- Hospital Santo Antônio, Praça Dr. Márcio Carvalho Lopes Filho, Curvelo, 501, Brazil
| | | | | | | | - Marcos André Gonçalves
- Computer Science Department, Universidade Federal de Minas Gerais, Av. Presidente Antônio Carlos, Belo Horizonte, 6627, Brazil
| | - Milena Soriano Marcolino
- Department of Internal Medicine, Medical School & Telehealth Center, University Hospital, Universidade Federal de Minas Gerais, Av. Professor Alfredo Balena, Belo Horizonte, 110, Brazil
- Institute for Health and Technology Assessment. R. Ramiro Barcelos, Porto Alegre, 2350, Brazil
| |
Collapse
|
2
|
Bali Swain R, Lin X, Wallentin FY. COVID-19 pandemic waves: Identification and interpretation of global data. Heliyon 2024; 10:e25090. [PMID: 38327425 PMCID: PMC10847870 DOI: 10.1016/j.heliyon.2024.e25090] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/01/2023] [Revised: 01/04/2024] [Accepted: 01/19/2024] [Indexed: 02/09/2024] Open
Abstract
The mention of the COVID-19 waves is as prevalent as the pandemic itself. Identifying the beginning and end of the wave is critical to evaluating the impact of various COVID-19 variants and the different pharmaceutical and non-pharmaceutical (including economic, health and social, etc.) interventions. We demonstrate a scientifically robust method to identify COVID-19 waves and the breaking points at which they begin and end from January 2020 to June 2021. Employing the Break Least Square method, we determine the significance of COVID-19 waves for global-, regional-, and country-level data. The results show that the method works efficiently in detecting different breaking points. Identifying these breaking points is critical for evaluating the impact of the economic, health, social and other welfare interventions implemented during the pandemic crisis. Employing our method with high frequency data effectively determines the start and end points of the COVID-19 wave(s). Identifying waves at the country level is more relevant than at the global or regional levels. Our research results evidenced that the COVID-19 wave takes about 48 days on average to subside once it begins, irrespective of the circumstances.
Collapse
Affiliation(s)
- Ranjula Bali Swain
- Department of Economics, Södertörn University, 141 89 Huddinge, Stockholm, Sweden
- Center for Sustainability Research (SIR), Stockholm School of Economics, Box 6501, SE-11383, Stockholm, Sweden
| | - Xiang Lin
- Department of Economics, Södertörn University, 141 89 Huddinge, Stockholm, Sweden
| | | |
Collapse
|
3
|
Tiwari S, Petrov AN, Golosov N, Devlin M, Welford M, DeGroote J, Degai T, Ksenofontov S. Regional geographies and public health lessons of the COVID-19 pandemic in the Arctic. Front Public Health 2024; 11:1324105. [PMID: 38259778 PMCID: PMC10801898 DOI: 10.3389/fpubh.2023.1324105] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/19/2023] [Accepted: 12/15/2023] [Indexed: 01/24/2024] Open
Abstract
Objectives This study examines the COVID-19 pandemic's spatiotemporal dynamics in 52 sub-regions in eight Arctic states. This study further investigates the potential impact of early vaccination coverage on subsequent COVID-19 outcomes within these regions, potentially revealing public health insights of global significance. Methods We assessed the outcomes of the COVID-19 pandemic in Arctic sub-regions using three key epidemiological variables: confirmed cases, confirmed deaths, and case fatality ratio (CFR), along with vaccination rates to evaluate the effectiveness of the early vaccination campaign on the later dynamics of COVID-19 outcomes in these regions. Results From February 2020 to February 2023, the Arctic experienced five distinct waves of COVID-19 infections and fatalities. However, most Arctic regions consistently maintained Case Fatality Ratios (CFRs) below their respective national levels throughout these waves. Further, the regression analysis indicated that the impact of initial vaccination coverage on subsequent cumulative mortality rates and Case Fatality Ratio (CFR) was inverse and statistically significant. A common trend was the delayed onset of the pandemic in the Arctic due to its remoteness. A few regions, including Greenland, Iceland, the Faroe Islands, Northern Canada, Finland, and Norway, experienced isolated spikes in cases at the beginning of the pandemic with minimal or no fatalities. In contrast, Alaska, Northern Sweden, and Russia had generally high death rates, with surges in cases and fatalities. Conclusion Analyzing COVID-19 data from 52 Arctic subregions shows significant spatial and temporal variations in the pandemic's severity. Greenland, Iceland, the Faroe Islands, Northern Canada, Finland, and Norway exemplify successful pandemic management models characterized by low cases and deaths. These outcomes can be attributed to successful vaccination campaigns, and proactive public health initiatives along the delayed onset of the pandemic, which reduced the impact of COVID-19, given structural and population vulnerabilities. Thus, the Arctic experience of COVID-19 informs preparedness for future pandemic-like public health emergencies in remote regions and marginalized communities worldwide that share similar contexts.
Collapse
Affiliation(s)
- Sweta Tiwari
- ARCTICenter, University of Northern Iowa, Cedar Falls, IA, United States
- Department of Geography, University of Northern Iowa, Cedar Falls, IA, United States
| | - Andrey N. Petrov
- ARCTICenter, University of Northern Iowa, Cedar Falls, IA, United States
- Department of Geography, University of Northern Iowa, Cedar Falls, IA, United States
| | - Nikolay Golosov
- Department of Geography, Pennsylvania State University, University Park, PA, United States
| | - Michele Devlin
- United States Army War College, Carlisle, PA, United States
| | - Mark Welford
- Department of Geography, University of Northern Iowa, Cedar Falls, IA, United States
| | - John DeGroote
- Department of Geography, University of Northern Iowa, Cedar Falls, IA, United States
| | - Tatiana Degai
- ARCTICenter, University of Northern Iowa, Cedar Falls, IA, United States
- Department of Anthropology, University of Victoria, Victoria, BC, Canada
| | - Stanislav Ksenofontov
- ARCTICenter, University of Northern Iowa, Cedar Falls, IA, United States
- Department of Geography, University of Northern Iowa, Cedar Falls, IA, United States
| |
Collapse
|