Applications of machine learning in decision analysis for dose management for dofetilide
Andrew E Levy1, Minakshi Biswas1, Rachel Weber2
1Division of Cardiology, University of Colorado Anschutz Medical Campus, Aurora, CO, United States of America.
Dose adjustments significantly impact dofetilide initiation success. A new reinforcement learning algorithm accurately predicts dosing decisions, aiming to improve patient outcomes and reduce toxicity risks.
Area of Science:
- Cardiology
- Pharmacology
- Artificial Intelligence in Medicine
Background:
- Dofetilide initiation requires FDA-mandated telemetry monitoring due to toxicity risks.
- Real-world dofetilide dosing strategies vary despite existing algorithms.
Purpose of the Study:
- To identify predictors of successful dofetilide loading.
- To develop a predictive model for dofetilide dosing decisions.
Main Methods:
- Analysis of clinical data from 354 patients in the Antiarrhythmic Drug Genetic (AADGEN) study.
- Utilized logistic regression, principal component analysis, cluster analysis, and reinforcement learning.
- Trained a reinforcement learning model on 80% of dosing decisions and tested on 20%.
Main Results:
- Starting dose of 500 mcg and sinus rhythm predicted successful dofetilide loading.
- Dose adjustments and coronary artery disease history predicted unsuccessful loading.
- Reinforcement learning model achieved 96.1% accuracy in predicting dosing decisions.
Conclusions:
- Dose adjustments are critical for successful dofetilide initiation.
- An unsupervised learning-informed reinforcement learning algorithm accurately predicts dosing decisions.
- This algorithm shows promise as a prospective data-driven decision aid for clinicians.
Related Concept Videos
Dose Size and Dosing Frequency: Determination Methods
Determination of Multiple Dosing Parameters: Loading and Maintenance Doses
Determination of Multiple Dosing Parameters: Steady-State, Minimum and Maximum Concentrations
Drug Dosing in Renal Diseases: Dose Adjustments Based on Drug Clearance and Elimination Rate Constant
Pharmacokinetic Models: Comparison and Selection Criterion
Physiological models take a detailed approach by considering specific molecular processes. They can predict drug distribution, metabolism, and elimination changes, providing a comprehensive understanding of how drugs interact with the body.
Dosage Regimens: Designs and Approaches


