Comprehensive data optimization and risk prediction framework: machine learning methods for inflammatory bowel...

Area of Science:

• Microbiome Research
• Computational Biology
• Disease Prediction

Background:

• Inflammatory Bowel Disease (IBD) prevalence is rising, necessitating early detection for better patient outcomes.
• Gut microbiome alterations are strongly associated with IBD onset and progression.
• Existing gut microbiome data challenges include missing values and high dimensionality, hindering predictive model accuracy.

Purpose of the Study:

• To develop a comprehensive data optimization and risk prediction framework (CDORPF) for predicting IBD risk.
• To address data challenges in gut microbiome datasets for improved IBD prediction.
• To aid in the early diagnosis of IBD through microbiome analysis.

Main Methods:

• The CDORPF framework integrates data optimization and risk prediction modules.
• Data optimization involves triple optimization imputation (TOI) for missing values and importance-weighted variational autoencoder (IWVAE) for dimensionality reduction.
• Risk prediction utilizes a random forest (RF) model with hyperparameters optimized by the improved aquila optimizer (IAO).

Main Results:

• The proposed CDORPF framework demonstrated high performance on IBD-related gut microbiome datasets.
• Achieved classification accuracy, recall, and F1 scores exceeding 0.9.
• Outperformed existing models in predicting IBD onset risk.

Conclusions:

• The CDORPF framework offers a robust solution for analyzing complex gut microbiome data.
• It effectively improves the accuracy and efficiency of IBD risk prediction.
• Serves as a valuable tool for early IBD diagnosis and management.