Live demo · Machine learning

Adverse event triage, scored in real time.

A working example from the portfolio: an XGBoost classifier that estimates the probability a pharmacovigilance case will be triaged as Serious, blending structured patient data with TF-IDF features from reported reactions.

Patient inputs
Risk score
Submit the form to generate a probability.

Trained with XGBoost on structured patient features and TF-IDF vectors of free-text adverse-event narratives. The score is the model's probability that a case would be triaged as Serious.

Under the hood

How the score is calculated.

Model
XGBoost gradient-boosted trees
Text features
TF-IDF over reaction narratives
Structured features
Age, weight, sex, case priority
Output
Probability of class Serious (0–1)