vp298/mbfbert-drug-adr

MBFBert Drug-ADR Prediction Model

This model predicts Adverse Drug Reactions (ADRs) for pharmaceutical compounds using dual MLP encoders.

Model Description

• Architecture: Dual MLP encoders (drug encoder + ADR encoder)
• Task: Drug-ADR association prediction
• Training Data: Drug and ADR embeddings
• Similarity Metric: Cosine similarity

Usage

from mbfbert import DrugADRPipeline

# Load model from Hugging Face
pipeline = DrugADRPipeline.from_pretrained("VinodPetluri/mbfbert-drug-adr")

# Make predictions with threshold
predictions = pipeline(drug_vector, top_k=50, threshold=0.5)

for adr, score in predictions:
    print(f"{adr}: {score:.4f}")