Developing an AI-based prediction model for anaphylactic shock from injection drugs using Japanese real-world data...

Area of Science:

• Pharmacovigilance
• Artificial Intelligence in Drug Safety
• Computational Chemistry

Background:

• Anaphylactic shock is a severe adverse drug reaction.
• Predicting drugs that induce anaphylactic shock is crucial for patient safety.

Purpose of the Study:

• To develop an AI-based prediction model for injection drugs causing anaphylactic shock.
• To utilize Japanese Real-World Data (JADER database) and chemical structure analysis.

Main Methods:

• Extracted adverse drug reaction data from JADER (April 2004-December 2020).
• Selected drugs with "anaphylactic shock" as an adverse reaction.
• Employed logistic regression, LASSO, XGBoost, RF, SVM, and NNW models using chemical properties and structural similarities.
• Applied principal component analysis for dimension reduction and cross-validation for optimization.

Main Results:

• Analyzed 227 injection drugs associated with anaphylactic shock.
• Evaluated model performance using AUCs, sensitivity, and specificity.
• Identified mepivacaine structural similarity as a key predictor in the final ensemble model.

Conclusions:

• Successfully developed an AI model to predict anaphylactic shock-inducing injection drugs.
• The model shows potential for enhancing drug safety assessments and anaphylactic shock risk evaluation.