Correlation of Lung Lesion Volume Measurement Using Artificial Intelligence and Prognosis of Patients with Severe Coronavirus Disease 2019 Infection

Abstract

Objective: To analyze the correlation between lung lesion volume and associated underlying diseases and prognosis of patients with severe coronavirus disease infection (COVID-19). Method: We reviewed 136 patients with severe COVID-19 in our hospital from December 8, 2022 to January 31, 2023. We measured the volume of lung lesions using artificial intelligence (AI), collected concomitant basic disease data and laboratory tests, and analyzed their impact on the prognosis of severe COVID-19. Results: The difference in the different prognoses of severe COVID-19, such as age, hypoproteinemia, stroke, lactate dehydrogenase, blood urea nitrogen (BUN), prothrombin time, albumin, leukocyte, lymphocyte ratio, neutrophil ratio, C-reactive protein, D-dimer, total lung lesion volume (TLLV), and percentage of total lung lesion volume (PTLLV), between the two groups was significant. Age, TLLV, PTLLV, BUN, and white blood cells were positively correlated with poor prognosis, while albumin was negatively correlated with poor prognosis. Conclusion: The older, the larger TLLV and PTLLV are, the more likely the patients with severe COVID-19 will have poor prognosis. The increase in indicators, such as BUN and white blood cells, and decrease in albumin are the risk factors for poor prognosis of the patients with severe COVID-19.

Keywords

• artificial intelligence
• covid-19
• lung lesion volume