Precision Surgery for Pediatric Hydrocephalus: VPS vs. ETV With ML-Guided Prediction

Individualized Treatment Strategies and Long-term Prognosis of Congenital Hydrocephalus: An Integrated Analysis Based on Multicenter Data and Machine Learning

Study type: Multicenter retrospective cohort study with prospective validation Primary purpose: To determine whether an individualized, imaging-guided treatment algorithm Ventriculoperitoneal shunt (VPS) vs endoscopic third ventriculostomy (ETV) improves 2-year neurodevelopmental outcomes in children <18 y with congenital hydrocephalus.

Main questions

• Does ETV produce higher 6-month surgical success and lower 2-year re-intervention rates than VPS in prespecified subgroups (age ≥3 y, obstructive hydrocephalus, normal basal cisterns)?
• Does a machine-learning model (ETV-PS) using pre-operative MRI features accurately predict ETV success (AUC ≥0.80) and thereby reduce unnecessary re-operations?
• Does early, frequent programmable-valve pressure adjustment after VPS decrease over-shunting headaches and improve 2-year cognitive scores compared with standard, infrequent adjustment?

Comparison: ETV group vs. VPS group (1:1 propensity-matched); within VPS cohort, frequent (≥3 adjustments in first 6 mo) vs. infrequent (<3) pressure-tuning arms.

Participants will

• Provide pre-operative clinical data and MRI/CT imaging.
• Undergo either VPS or ETV as clinically indicated; 320 VPS recipients receive programmable valves with protocol-driven pressure logs.
• Return for standardized neurodevelopmental testing Children's Memory Scale (CMS), Wechsler Intelligence Scale for Children(WISC), Pediatric Quality of Life Inventory(PedsQL) and imaging at 6 mo, 1 y and 2 y; valve adjustments tracked electronically.

Inclusion Criteria

• Age 0-17.99 years at surgery
• Congenital hydrocephalus confirmed by clinical and MRI/CT findings
• First surgical treatment: either VPS or ETV
• Complete pre-operative MRI/CT and ≥ 2-year follow-up data available

Exclusion Criteria

• Secondary hydrocephalus (tumor, infection, trauma)
• Severe comorbidities affecting neuro-developmental assessment (e.g., major congenital heart disease, genetic-metabolic disorders)
• Incomplete baseline imaging or follow-up < 2 years