Antiviral Effectiveness, Clinical Outcomes, and Artificial Intelligence Imaging Analysis for Hospitalized COVID-19 Patients Receiving Antivirals

Yuan Gao,Yixi Dong,Qiushi Bu,Zhijie Gong,Wei Wang,Zhongkai Zhou,Yunyi Gao,Liwei Liu,Menghua Wu,Jiaying Zhang,Lianchun Liang,Hongjun Li,Mengxi Jiang,Zujin Luo,Yingmin Ma,Xinyu Zhang,Zhongjie Hu
DOI: https://doi.org/10.1111/irv.70006
Abstract:Introduction: There is still a lack of clinical evidence comprehensively evaluating the effectiveness of antiviral treatments for COVID-19 hospitalized patients. Methods: A retrospective cohort study was conducted at Beijing You'An Hospital, focusing on patients treated with nirmatrelvir/ritonavir or azvudine. The study employed a tripartite analysis-viral dynamics, survival curve analysis, and AI-based radiological analysis of pulmonary CT images-aiming to assess the severity of pneumonia. Results: Of 370 patients treated with either nirmatrelvir/ritonavir or azvudine as monotherapy, those in the nirmatrelvir/ritonavir group experienced faster viral clearance than those treated with azvudine (5.4 days vs. 8.4 days, p < 0.001). No significant differences were observed in the survival curves between the two drug groups. AI-based radiological analysis revealed that patients in the nirmatrelvir group had more severe pneumonia conditions (infection ratio is 11.1 vs. 5.35, p = 0.007). Patients with an infection ratio higher than 9.2 had nearly three times the mortality rate compared to those with an infection ratio lower than 9.2. Conclusions: Our study suggests that in real-world studies regarding hospitalized patients with COVID-19 pneumonia, the antiviral effect of nirmatrelvir/ritonavir is significantly superior to azvudine, but the choice of antiviral agents is not necessarily linked to clinical outcomes; the severity of pneumonia at admission is the most important factor to determine prognosis. Additionally, our findings indicate that pulmonary AI imaging analysis can be a powerful tool for predicting patient prognosis and guiding clinical decision-making.
What problem does this paper attempt to address?