Monkeypox diagnosis based on probabilistic K-nearest neighbors (PKNN) algorithm.

Area of Science:

• Artificial Intelligence in Medicine
• Medical Image Analysis
• Epidemiology

Background:

• Monkeypox resurgence presents diagnostic challenges.
• Existing diagnostic methods may lack speed and accuracy.
• Need for advanced computational approaches in disease detection.

Purpose of the Study:

• To present a novel artificial intelligence approach for monkeypox diagnosis, termed Effective Monkeypox Diagnosis Strategy (EMDS).
• To improve the accuracy and efficiency of monkeypox detection using machine learning.
• To introduce new methodologies for data preprocessing and classification in medical diagnostics.

Main Methods:

• Developed EMDS with two stages: Pre-Processing Phase (PP) and Monkeypox Diagnosing Phase (MDP).
• Utilized GoogleNet for feature extraction and Leopard Seal Optimization (LSO) for feature selection.
• Introduced an outlier rejection methodology (ORM) using interquartile range (IQR) and a Probabilistic K-Nearest Neighbors (PKNN) classifier.

Main Results:

• EMDS achieved 99% accuracy in diagnosing monkeypox across two public datasets (MSID and MSLD).
• The proposed PKNN classifier and ORM individually showed significant performance improvements.
• EMDS demonstrated superior performance compared to recent monkeypox identification strategies, with high precision, recall, and minimal diagnosis time.

Conclusions:

• The EMDS offers a highly accurate and efficient AI-based solution for monkeypox diagnosis.
• The novel ORM and PKNN components significantly contribute to the overall diagnostic performance.
• This AI strategy holds promise for rapid and reliable detection of monkeypox outbreaks.