Diagnosing Cancer Using IOT and Machine Learning Methods.

Area of Science:

• Oncology
• Bioinformatics
• Computational Biology

Background:

• Breast cancer is the most prevalent cancer affecting women globally, necessitating advanced diagnostic tools.
• Microarray technology generates large datasets, enabling sophisticated computational analysis for disease detection.

Purpose of the Study:

• To investigate the efficacy of machine learning algorithms for classifying breast cancer using microarray data.
• To evaluate different feature reduction techniques and their impact on classification accuracy.

Main Methods:

• Utilized Python to implement machine learning algorithms (Logistic Regression, Random Forest, SVM, Adaboost, Gradient Boosting Machine, MLP) on two distinct microarray datasets.
• Compared classification performance with and without feature reduction strategies.
• Analyzed the impact of deep learning model depth on diagnostic accuracy.

Main Results:

• Support Vector Machines (SVM) demonstrated the highest accuracy, reaching 99.23% on the first dataset and 88.82% on the second.
• Logistic Regression yielded 90.23% accuracy before feature reduction.
• Increasing deep learning model layers beyond a certain point did not enhance classification accuracy, with maximums of 97.69% and 68.72%.

Conclusions:

• Machine learning, particularly SVM, shows significant potential for accurate breast cancer diagnosis from microarray data.
• Feature reduction techniques may influence, but do not universally improve, classification accuracy.
• Deep learning model complexity requires careful consideration, as increased depth does not guarantee better performance.