Classification of pigmented lesion using novel k-nearest neighbor comparing with logistic regression algorithm for better accuracy

• Main Authors: Sha, S. Ashwin, Saravanan, M. S.
• Format: Conference Proceeding
• Language: English
• Subjects: Accuracy; Algorithms; Cancer; Classification; Confidence intervals; Internet resources; Machine learning; Regression analysis; Skin cancer; Statistical analysis
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0197384

Recent technology breakthroughs have made it possible to automatically detect skin cancer using machine learning algorithms. Nonmelanoma skin cancers of the head and neck may now be diagnosed precisely and promptly thanks to the use of machine learning (NMSC). The researchers set out to find real-world medicinal uses for machine learning. K-Nearest Neighbor (KNN) and Logistic Regression were used as unique classification methods in this research with two distinct sample sizes (LR). Groups 1 and 2 of the KNN and LR algorithms had g-power values of 80%, and the squamous cell cancer images were collected from a wide range of internet resources using up-to-date research outcomes, a 0.05 percent threshold, a 95% confidence interval for the mean, and a standard deviation. Researchers often resort to methods that depend on machine learning because of how challenging it is to diagnose these illnesses. The KNN method achieved 86.11 percent accuracy whereas the LR algorithm only achieved 82 percent accuracy, with a significant value of 0.035 (p0.05) and a 95 percent confidence interval. Recommendations for POC skin cancer prevention and early detection are guided by a thorough analysis of the available data using the Machine Learning Approach.