Open Spina Bifida Survival Studio
Landmark survival prediction across 5 life stages — with explainable AI.
This tool predicts long-term survival probability for patients with open spina bifida (myelomeningocele) using landmark survival analysis, a machine learning approach where the model updates as a patient ages and more clinical data becomes available, selecting a more detailed predictor trained on that life stage.
Five Random Survival Forest models are available (at birth, age 5, 25, 35, and 40), each incorporating progressively more features (birth neurology, childhood outcomes, adolescent mobility, adult independence, etc.). Select the patient's current age group below to see which features are relevant.
After prediction, you can run a SHAP (SHapley Additive exPlanations) analysis to see which specific features are driving this patient's risk up or down, providing interpretable, per-patient explainability.
Based on the Cambridge cohort of open spina bifida (n=117, born 1963-1971, followed to age 50). Random Survival Forest with 3-fold cross-validation. For research demonstration only, not for clinical decision-making.
First run may take up to 60 seconds (server cold start). For research demonstration only. Not for clinical decision-making.