Artificial Intelligence Algorithm Based on Genetics to Predict Responses to Interferon-Beta Treatment in Multiple...

Area of Science:

• Neuroimmunology
• Artificial Intelligence in Medicine
• Computational Biology

Background:

• Multiple sclerosis (MS) is a CNS inflammatory disease affecting millions, with treatment response influenced by genetic factors.
• Interferon-beta (IFN-β) is a common MS therapy, but 30-50% of patients show inadequate response due to genetic variability.
• Current machine learning methods for classifying treatment response in MS have limitations in handling complex datasets.

Purpose of the Study:

• To develop and evaluate an AI-driven fuzzy system for improved classification of IFN-β response in MS patients.
• To utilize a genetic algorithm (GA) for optimizing deep learning (DL) model hyperparameters for predicting therapy benefits.
• To enhance the accuracy of identifying MS patients likely to benefit from IFN-β treatment.

Main Methods:

• An AI algorithm combining a fuzzy system (FS) with neurologist input was developed to classify IFN-β response.
• A genetic algorithm (GA) was employed to optimize hyperparameters for a deep learning (DL) model trained on genetic biomarkers.
• Performance was compared against conventional methods like hierarchical clustering and standard multi-layer perceptron (MLP) tuning.

Main Results:

• The fuzzy system achieved 80% classification efficiency, outperforming hierarchical clustering (64%).
• A GA-optimized artificial neural network (ANN) model demonstrated high accuracy (0.8-1.0).
• The GA-optimized ANN surpassed the conventionally tuned MLP (0.6-0.8 accuracy).

Conclusions:

• AI-driven fuzzy systems offer a more effective approach for classifying complex biological data in MS.
• Genetic algorithm optimization significantly enhances the predictive performance of deep learning models for MS treatment response.
• These advanced computational methods hold promise for personalized medicine in multiple sclerosis management.