Data-Driven Modeling for Precision Medicine in Pediatric Acute Liver Failure

Area of Science:

• Pediatric Hepatology
• Translational Medicine
• Systems Biology

Background:

• Early outcome biomarkers for Pediatric Acute Liver Failure (PALF) are currently lacking.
• This absence complicates critical medical and liver transplant decisions for affected children.
• Understanding dynamic inflammatory processes may offer insights into PALF prognosis.

Purpose of the Study:

• To define dynamic interactions among circulating inflammatory mediators in PALF patients.
• To identify patterns associated with different patient outcomes (spontaneous survival, non-survival, liver transplant).
• To explore the potential of these interactions as predictive biomarkers.

Main Methods:

• Serum samples from 101 PALF participants were collected within 7 days of enrollment.
• 27 inflammatory mediators were assayed.
• Data-driven algorithms, including Dynamic Bayesian Network inference, were used to analyze mediator interrelations and network dynamics over time.

Main Results:

• A common network motif with HMGB1 as a central node was identified across all patient subgroups.
• Dynamic network connectivity differed between spontaneous survivors (S) and liver transplant (LTx) patients compared to non-survivors (NS).
• A Dynamic Robustness Index effectively differentiated between S, NS, and LTx subgroups, with higher interconnectedness observed in NS patients that increased over time.

Conclusions:

• Increasing inflammatory network connectivity in PALF is associated with non-survival.
• Dynamic network analysis of inflammatory mediators shows promise for stratifying PALF patient outcomes.
• These findings may lead to improved clinical decision-making and patient management in PALF.