On the relationship between energy-related plants and oncological cases in basilicata (Italy) using soft computing methods

In Basilicata (Southern Italy), in areas around energy-related plants, including oil extraction sites, oil refineries, and underground gas storage plants, we consider a set of annual air quality measurements, the analysis of toxic substances emitted, and the percentage of tumours with respect the habitants. Artificial Neural Networks and Genetic Programming are then applied in order to assess the data correlation and to estimate the tumour percentage in the next years. The approach is tested using a tenfold cross validation methodology. Both the used soft computing methods show low error rates and high correlation measures. Furthermore the Genetic Programming evidences an explicit representation of the factors that favour the tumours. The results push the attention towards the prevention of potential health impacts among Basilicata residents living close to the plants.