MapReduce based big data framework using associative Kruskal poly Kernel classifier for diabetic disease prediction.

Area of Science:

• Artificial Intelligence
• Machine Learning
• Big Data Analytics
• Computational Health

Background:

• Machine Learning (ML) algorithms are increasingly used for complex tasks like disease prediction.
• The growth of big data necessitates accelerated computation for effective ML applications in healthcare.
• Early disease prediction requires efficient computational methods to leverage ML's potential.

Purpose of the Study:

• To present a novel method, AKW-MRPK, for accelerating early disease prediction using ML on big data.
• To enhance the accuracy and speed of disease prognosis through optimized computational techniques.
• To demonstrate the effectiveness of parallelized polynomial kernels in healthcare data analysis.

Main Methods:

• Feature selection using the Associative Kruskal Wallis model to identify significant attributes.
• Parallelization of polynomial kernel vectors via MapReduce based on selected features.
• Implementation of the AKW-MRPK framework for disease prediction.

Main Results:

• The AKW-MRPK framework achieved up to 92% accuracy in early disease prediction.
• Computational time was reduced to 0.875 ms for 25 patients.
• Demonstrated superior speedup efficiency (1.9 ms using two nodes) compared to traditional methods.

Conclusions:

• The AKW-MRPK method effectively selects attributes and accelerates computations for improved disease prediction.
• Parallelizing polynomial kernels enhances both accuracy and speed in healthcare big data analysis.
• The proposed framework offers a significant advancement for early disease prognosis.