Figure 2

Correlation-based network analysis. The Pearson correlation threshold of 50% was used to show the network of all variables (a–c). More precisely, the Pearson correlation thresholds of 72% and 93% (d, e) were respectively selected to define the connection between survivors, non-survivors and blood parameters. The Pearson correlation thresholds of 79% and 97% were respectively chosen to assess the relationship between survivors, non-survivors and (f) risk factors or (g) complications. Circles of the network indicate the blood parameters (a, d, e), risk factors (b, f) and complications (c, g). The size of the node reflects the degree of connectivity of the node and the edges display the relationship between the two variables. The thicker edges reveal higher correlations between variables. Nodes with more links are close to each other. Network analysis and visualization was carried out using PAST and Fruchterman-Reingold algorithm or Circular algorithm as a force-directed layout algorithm. Abbreviations in panels (a), (d), and (e): Alb, albumin; HBG, hemoglobin; NEU, neutrophil; PLT, platelet; LYM, lymphocyte; WBC, white blood cells, PCT, procalcitonin; GGT, gamma-glutamyl transferase; CRP, C-reactive protein, CK, creatine kinase; Creat: creatinine, BUN, blood urea nitrogen; Bili, total bilirubin; AST, aspartate aminotransferase; ALT, alanine aminotransferase. Abbreviations in panels (b) and (f): BMI, body mass index; Time to H, time from symptoms appearance to hospitalization; Renal, renal disease; Cereb, cerebrovascular disease; Liver, liver disease; COPD, chronic obstructive pulmonary disease; Cardio, cardiovascular disease. Abbreviations in panels (c) and (g): Fail, failure; Res, respiratory; Liver, liver dysfunction; Sec, secondary; Kidney, acute kidney injury; Cardiac, acute kidney injury.