Fig. 3: MPSE enables automated WGS candidate identification within the first 24 h in the NICU.

Panel A shows MPSE score distributions across the first 96 h in the NICU for diagnostic (red) and sequenced but non-diagnostic (blue) patients as well as unsequenced NICU patients (green). Diagnostic and non-diagnostic sequenced patients had significantly higher MPSE scores than unsequenced patients beginning 0–24 h after admission. Diagnostic patients had significantly higher MPSE scores than non-diagnostic patients beginning at 48 h post-admission. Boxplot comparison significance levels: *** (p < 1e−5); * (p < 0.05). Panel B shows MPSE score trajectories for these groups across the first 30 days in the NICU. Solid lines show the mean MPSE score per group and the shaded regions cover one standard deviation from each mean. Panel C shows the probability of at-risk patients being classified as a WGS candidate by MPSE (i.e., MPSE score >2 standard deviations above the unsequenced mean score) as a function of time. Cox proportional hazards regression analysis confirmed the significantly increased rate of MPSE candidate selection for diagnostic and non-diagnostic patients selected for WGS compared to unsequenced patients, with hazards ratios of 18.8 (95% CI 11.6–30.6; p = 1.9e−32) and 9.8 (95% CI 6.2–15.3; p = 2.8e−23) respectively. Diagnostic patients were also selected by MPSE at a higher rate than sequenced but non-diagnostic patients (HR = 2.0; 95% CI 1.1–3.9; p = 0.03). Figure generated with R ggplot2 software.