On the use of mining techniques to analyse human papilloma virus dataset

Human papilloma virus (HPV) is a type of infection that can be pathogenic for the human. In many cases, HPV infection can produce precancerous lesions in skin and mucous membranes in the body causing genital warts and cervical cancer. The idea of the proposed contribution is to develop a dedicate framework to support clinical activity in HPV treatment.Data mining techniques have been proposed to analyze HPV data coming from the microbiology unit of Magna Graecia University. Bayesian and k-Nearest Neighbor (k-NN) algorithms have been applied to HPV data stored in a dataset aiming to extract relevant clinical information useful to support physician and biologist in HPV evaluation. Results show that k-NN algorithm allows a more accurate prediction in the gender affected by the infection compared to Bayesian algorithm. Another relevant result is that the high-risk type of virus HPV16 represents the most common genotype for male and female. Finally, the heat map method has been applied to observe the relevant correlation between HPV genotypes and their relative risk level.