Klusterirakenteista syöpäilmaantuvuus ja -kuolleisuusdatassa yli ajan Suomessa

The cancer burden is increasing globally. In 2022, nearly 20 million new cancer cases were diagnosed, and by 2050, this figure has been estimated to increase by 77%, or over 35 million new cases. Part of this development can be explained by the expected increase in population during the same time frame from 8.0 billion to 9.7 billion. In addition, the share of people aged 65 years and above is expected to increase from 10% to 16%, and aging, after all, is among the main risk factors causing cancer. However, our current way of living and environment are associated with many other risk factors causing cancer, including an unbalanced diet, obesity, physical inactivity, smoking, alcohol use, microplastics, and xenoestrogens. In this thesis, cluster structures of cancer incidence and mortality data from 1962 to 2022 in Finland are identified and analyzed. The analysis is divided between females and males, and different age groups, ranging from 20-29 to 70-79 years. Both unstandardized and standardized data is used. To identify the cluster structures, an agglomerative hierarchical clustering algorithm is utilized, combined with a tailored proximity measure and the average linkage method. After employing the clustering algorithm, the resulting cluster structures are described, differences in the cluster structures between different subgroups determined by age and gender are evaluated, and whether hormone-related cancers emerge in the same cluster due to the Western lifestyle is considered. To support the analysis, a description of the Western lifestyle, as well as the associations between its components and cancer, are also provided as part of the thesis. In terms of results of the thesis, it was observed that in many cases there is one large cluster containing many different cancers, while in the remaining clusters, there is often only one, sometimes two cancers per cluster. As expected, it appeared that differences in the scale affected the resulting clustering structures a bit in the case of unstandardized data. It was also discovered that hormone-related cancers, such as breast, cervical, and prostate cancer often formed clusters of their own, contrary to forming joint clusters with other hormone-related cancers. In addition, lung and tracheal cancer was observed in many cases, both among females and males, form a cluster of its own, suggesting possible changes in smoking behavior.