Predicting Genetic Disorder and Types of Disorder Using Chain Classifier Approach.

Area of Science:

• Genomics
• Bioinformatics
• Artificial Intelligence

Background:

• Genetic disorders stem from DNA mutations, leading to severe conditions like Alzheimer's and cancer.
• Artificial intelligence (AI) shows promise in disease prediction and prognosis.
• Early prediction of genetic disorders using genomic data is crucial for timely intervention.

Purpose of the Study:

• To develop an effective AI-based method for early prediction of genetic disorders.
• To address the multi-label, multi-class nature of genetic disorder prediction.
• To enhance prediction accuracy and computational efficiency compared to existing methods.

Main Methods:

• A novel feature engineering approach combining class probabilities from Extra Trees (ET) and Random Forests (RF).
• Utilizing a classifier chain approach for multi-label prediction, where classifiers leverage preceding predictions.
• Employing macro accuracy, Hamming loss, and alpha-evaluation score for performance assessment.

Main Results:

• Extreme Gradient Boosting (XGB) achieved the highest performance.
• XGB demonstrated a 92% alpha-evaluation score and 84% macro accuracy.
• The proposed XGB model outperformed state-of-the-art methods in accuracy and computational efficiency.

Conclusions:

• The novel feature engineering and classifier chain approach, particularly with XGB, is highly effective for genetic disorder prediction.
• This AI-driven strategy offers a significant advancement for early detection and management of genetic diseases.
• The method provides a computationally efficient and accurate solution for complex genetic disorder prediction.