tetano
Editor, Senior Moderator
PeerJ
. 2026 Feb 10:14:e20795.
doi: 10.7717/peerj.20795. eCollection 2026.
Study on the mortality risk and predictive model for COVID-19 inpatients with pneumonia manifestations
Zhi Li[SUP] #[/SUP][SUP] 1 [/SUP], Jiamin Liang[SUP] #[/SUP][SUP] 1 [/SUP], Katie Lu[SUP] 2 [/SUP], Shuyu Tang[SUP] 3 [/SUP], Jinyi Huang[SUP] 1 [/SUP], Jinrong Zhang[SUP] 1 [/SUP], Jianjun Zou[SUP] 1 [/SUP], Dongsheng Huang[SUP] 4 [/SUP], Chenli Xie[SUP] 5 [/SUP], Linglong Zeng[SUP] 1 [/SUP], Zhiwei Wang[SUP] 6 [/SUP], Yibin Deng[SUP] 7 [/SUP], Jiachun Lu[SUP] 1 [/SUP]
Affiliations
Background: In 2020, COVID-19 posed a major threat to global public health in a remarkably short period. Although the WHO declared an end to the emergency phase in May 2023, a considerable proportion of recovered cases experience medium- and long-term effects, which pose ongoing health challenges to society. Therefore, it remains necessary to conduct relevant research in the post-epidemic era to explore the risk factors for death in COVID-19 inpatients.
Methods: We determined the mortality of COVID-19 inpatients with pneumonia manifestations through one-year follow-up, utilizing real-world data from three medical centers. Clinical characteristics associated with mortality risk were analyzed by logistic regression. Then, the dataset was randomly partitioned into three sets at a ratio of 4:2:4. Three machine learning algorithms were employed to develop and validate a mortality risk predictive model for COVID-19 inpatients, and a web-based visualization tool was created.
Results: There were 100 fatalities among the 1,693 samples included in this study. Meanwhile, we identified 37 factors correlated with increased mortality risk in COVID-19 inpatients with pneumonia manifestations. Ultimately, we developed a mortality risk predictive model using the random forest algorithm, which demonstrated superior predictive performance (AUC=0.907, 95% CI=0.849-0.957).
Conclusions: This study reports a mortality rate of 5.9% for COVID-19 inpatients with pneumonia manifestations. The high-performance mortality risk prediction model obtained in this study provides important practical guidance for monitoring the mortality risks of COVID-19 inpatients with pneumonia manifestations.
Keywords: COVID-19; In-patient; Machine learning; Mortality risk; Prediction model.
. 2026 Feb 10:14:e20795.
doi: 10.7717/peerj.20795. eCollection 2026.
Study on the mortality risk and predictive model for COVID-19 inpatients with pneumonia manifestations
Zhi Li[SUP] #[/SUP][SUP] 1 [/SUP], Jiamin Liang[SUP] #[/SUP][SUP] 1 [/SUP], Katie Lu[SUP] 2 [/SUP], Shuyu Tang[SUP] 3 [/SUP], Jinyi Huang[SUP] 1 [/SUP], Jinrong Zhang[SUP] 1 [/SUP], Jianjun Zou[SUP] 1 [/SUP], Dongsheng Huang[SUP] 4 [/SUP], Chenli Xie[SUP] 5 [/SUP], Linglong Zeng[SUP] 1 [/SUP], Zhiwei Wang[SUP] 6 [/SUP], Yibin Deng[SUP] 7 [/SUP], Jiachun Lu[SUP] 1 [/SUP]
Affiliations
- PMID: 41695709
- PMCID: PMC12903905
- DOI: 10.7717/peerj.20795
Background: In 2020, COVID-19 posed a major threat to global public health in a remarkably short period. Although the WHO declared an end to the emergency phase in May 2023, a considerable proportion of recovered cases experience medium- and long-term effects, which pose ongoing health challenges to society. Therefore, it remains necessary to conduct relevant research in the post-epidemic era to explore the risk factors for death in COVID-19 inpatients.
Methods: We determined the mortality of COVID-19 inpatients with pneumonia manifestations through one-year follow-up, utilizing real-world data from three medical centers. Clinical characteristics associated with mortality risk were analyzed by logistic regression. Then, the dataset was randomly partitioned into three sets at a ratio of 4:2:4. Three machine learning algorithms were employed to develop and validate a mortality risk predictive model for COVID-19 inpatients, and a web-based visualization tool was created.
Results: There were 100 fatalities among the 1,693 samples included in this study. Meanwhile, we identified 37 factors correlated with increased mortality risk in COVID-19 inpatients with pneumonia manifestations. Ultimately, we developed a mortality risk predictive model using the random forest algorithm, which demonstrated superior predictive performance (AUC=0.907, 95% CI=0.849-0.957).
Conclusions: This study reports a mortality rate of 5.9% for COVID-19 inpatients with pneumonia manifestations. The high-performance mortality risk prediction model obtained in this study provides important practical guidance for monitoring the mortality risks of COVID-19 inpatients with pneumonia manifestations.
Keywords: COVID-19; In-patient; Machine learning; Mortality risk; Prediction model.