In medicine, it is essential not only to accurately identify a specific disease based on some symptoms but also to plan a right treatment by doing what it is strictly required, timely and in due form. A primary example of the importance of early detection and rightly following treatment is cancer, one of the major public health problems worldwide [Bra18]. Every year, cancer not only entails social but also economic costs that cannot be easily covered by developing countries. Complications after cancer surgical may appear, and they are associated with high economic costs that can be alleviated if such complications can be timely forecasted. For these reasons, it is crucial to propose new study methods that clearly describe the disease at an early stage and reduce the costs associated with its treatment.
The emergence of next generation sequencing technologies has led to the development of new methods to study common alterations in diseases. The data obtained through these technologies is analysed with bioinformatics software tools to determine which genes are differentially expressed between the conditions (e.g., cancer vs no cancer). Techniques such as gene set analysis have emerged [Moo15] to evaluate the behaviour of functional groups of genes rather than individual genes. However, the formation of these groups requires prior biological knowledge that allows us to group genes with similar characteristics. Hence, even when this methodology is widely accepted in the field, it includes a major drawback: prior hypotheses need to be formulated with a background in the functional relationships between multiple genomic variables. New methodologies that avoid a bias in the searching process are required, enabling not only the looking for useful insights without any previous background in the field but also the analysis of high order relationships among variables as pattern mining does [Ven18]. Last but not least, the automatic diagnosis of different types of cancer, e.g., melanoma skin cancer, requires the development of highly efficient methods that ease the diagnostic at an early stage.
The aim is to overcome some of the existing limitations in early diagnosis through pattern mining techniques in the following ways:
- Application of classic and trending pattern mining techniques to extract useful knowledge in the study of the cancer. This would allow us to work without previous hypotheses, reaching unbiased conclusions that may be of great interest in the search for molecular markers which could be used for the early detection of cancer.
- Application of supervised descriptive pattern mining techniques to clearly differentiate genes associated to different types of cancer.
Furthermore, we want to continue the line.
- Application of new and improved Deep Learning methods to accurately identify melanoma skin cancer.
The research group has an important background in the pattern mining field [Ven16], the application of machine learning to cancer [Rey20] and melanoma detection [Per20], as well as supporting clinical decisions [Luq20].
[Per20] E. Perez, O. Reyes, S. Ventura. Convolutional neural networks for the automatic diagnosis of melanoma: An extensive experimental study. Medical image analysis 67, 101858. 2020. https://doi.org/10.1016/j.media.2020.101858
[Rey20] O. Reyes, E. Perez, J. Castaño, R. Luque, S. Ventura. A supervised machine learning-based methodology for analyzing dysregulation in splicing machinery: An application in cancer diagnosis. Art. Intell. in Med. 2020. DOI: doi.org/10.1016/j.artmed.2020.101950
[Bra18] F. Bray, J. Ferlay, I. Soerjomataram, R. L. Siegel, L. A. Torre, and A. Jemal. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: a cancer journal for clinicians 68, 6, 394–424. 2018.
[Luq20] C. Luque, J.M. Luna, S. Ventura. A semantically enriched text mining system for clinical decision support. Comp. Intelligence, 2020. https://doi.org/10.1111/coin.12322
[Ven16] S. Ventura, J. M. Luna: Pattern Mining with Evolutionary Algorithms. Springer 2016, ISBN 978-3-319-33857-6, pp. 1-190.
[Ven18] S. Ventura, J. M. Luna: Supervised Descriptive Pattern Mining. Springer 2018, ISBN 978-3-319-98139-0, pp. 1-185
[Moo15] M. A. Mooney, B. Wilmot, B. Gene set analysis: A step-by-step guide. American journal of medical genetics. Part B, Neuropsychiatric genetics: the official publication of the International Society of Psychiatric Genetics,168(7), 517-27. 2015.