Sensitivity Analysis by the PROMETHEE-GAIA method: Algorithms evaluation for COVID-19 prediction.

Area of Science:

• Computational epidemiology
• Machine learning applications in healthcare
• Decision analysis

Background:

• The global expansion of coronavirus necessitates advanced technological solutions for disease analysis and prediction.
• Machine learning (ML) algorithms offer potential for identifying and predicting infected patients.
• A structured approach is needed to evaluate and select the most effective ML algorithms for pandemic response.

Purpose of the Study:

• To analyze and compare machine learning algorithms for predicting and identifying COVID-19 infected patients.
• To apply a multicriteria decision-making approach to rank the performance of different prediction algorithms.
• To provide a transparent and effective method for selecting optimal algorithms to combat the COVID-19 pandemic.

Main Methods:

• Utilized the PROMETHEE-GAIA multicriteria method for algorithm analysis.
• Structured alternatives (ML algorithms) against a set of criteria to determine their importance.
• Performed sensitivity analysis using PROMETHEE I and II methods and the GAIA plan, implemented via Visual PROMETHEE software.

Main Results:

• The multicriteria analysis successfully structured and ranked the evaluated machine learning algorithms.
• Sensitivity analysis provided insights into algorithm performance under varying criteria.
• The PROMETHEE-GAIA method demonstrated effectiveness in transparently evaluating and selecting prediction algorithms.

Conclusions:

• The developed analysis model is effective for ranking machine learning algorithms for COVID-19 prediction.
• The PROMETHEE-GAIA method offers a robust framework for inter- and intra-criterion evaluation.
• This approach supports informed decision-making in selecting the best technological solutions for pandemic control.