Fig. 2: Group-level and patient-level estimates of cellular and molecular trajectories.
From: A patient-centric modeling framework captures recovery from SARS-CoV-2 infection

a, Recovery-group trajectories estimated by longitudinal mixed modeling for five parameters, one from each data type (cytokines, lymphocytes, kynurenine-pathway metabolites, lipoproteins and glycoproteins) over the first 7 weeks after symptom onset, with 95% confidence bands. All levels have been log-transformed and the gray bands correspond to the IQR of HCs’ levels. False discovery rate (FDR)-adjusted P values from LR tests for baseline and interaction effects are indicated (****P < 0.0001, ***P < 0.001, **P < 0.01 and *P < 0.05). b, Trajectories of the same parameters as in a, estimated by FPCA for three CovPs, one from each recovery group. Dashed curves delineate the 95% confidence bands. c, Comparison of the severity (lower diagonal part) and recovery scores (upper diagonal part) obtained from the six FPCA studies with Pearson simple correlation (‘Cor’) and partial correlation (‘pCor’); significant estimates at an FDR level of 5% are highlighted in bold. Severity classes: B, screening symptomatic; C, hospital no oxygen required; D, hospital supplemental oxygen; E, hospital assisted ventilation. d, Conditional independence networks for the severity and recovery scores; an edge is shown if the pair of score types is directly associated (partial correlation) at an FDR level of 5%, and the opacity and width of the edge is proportional to the strength of the association.