Integrated Bayesian and association-rules methods for autonomously orienting COVID-19 patients.

Area of Science:

• Artificial Intelligence
• Computational Biology
• Epidemiology

Background:

• The rapid global spread of coronavirus necessitates advanced tools for effective management.
• Accurate classification of COVID-19 symptom severity is crucial for appropriate treatment decisions.

Purpose of the Study:

• To develop a novel autonomous decision-making process for supporting COVID-19 management.
• To integrate Bayesian Networks for symptom classification and association rules for decision generation.

Main Methods:

• A Bayesian Network model was developed using the novel MIGT-SL algorithm for structure learning.
• Data analysis module classifies COVID-19 severity (mild, moderate, severe) using Bayesian Networks.
• An autonomous decision-making module employs association rules mining (FP-growth) for generating adequate decisions.

Main Results:

• The MIGT-SL algorithm accurately reconstructed network structures (74%-100%) in benchmark datasets.
• The Bayesian model achieved high classification accuracy: 96.15% (binary) and 94.77% (multi-class).
• The decision-making model demonstrated a high accuracy of 97.80% in proposing adequate decisions.

Conclusions:

• The proposed integrated system effectively supports decision-making in managing COVID-19.
• The novel algorithms enhance accuracy in both symptom classification and treatment decision generation.
• This AI-driven approach offers a promising tool for combating the impact of coronavirus infections.