The article outlines part of the analysis and modelling work related to Artificial Intelligence (AI) and Statistics led by the In Silico Oncology and In Silico Medicine Group in Athens, as part of the BOUNCE action.
Life after breast cancer treatment may be characterized by a number of complex issues involving biomedical, psychosocial and functional problems which in turn can affect the quality of life and the professional and other activities of the survivor. In order to optimize the resilience of the survivor in the personalised setting, various types of multidisciplinary support can be provided by specialised health professionals. An in depth understanding and quantification of the factors that influence resilience as well as the factor interdependences and interplays is a prerequisite for an efficient personalised tailoring of supportive actions and interventions.
To this end, both retrospective and prospective data being collected by the four clinical centres participating in the EU funded BOUNCE project (https://www.bounce-project.eu/) are analysed using methods of artificial intelligence (AI) and advanced statistics. It is noted that the BOUNCE project addresses the EU call SC1-PM-17-2017 entitled: “Personalised computer models and in-silico systems for well-being”. Within this context, the In Silico Oncology and In Silico Medicine Group, Institute of Communication and Computer Systems, National Technical University of Athens (https://www.in-silico-oncology.iccs.ntua.gr/), which leads the BOUNCE action on the development of the predictive breast cancer resilience computer models, has extensively analysed retrospective data collected by the Helsinki University Hospital, the Hebrew University of Jerusalem (HUJI), the Champalimaud Clinical Centre in Lisbon and the European Institute of Oncology in Milan, in close collaboration with the local team leaders and other clinical investigators.
Cross-sectional and longitudinal analyses have been performed in order to identify the effects of various factors on the status of the survivor at specific time points and to predict the trajectory of the survivor regarding various parameters for a given time interval. For example, in the case of retrospective data provided by HUJI, the steps taken so far include the following indicative ones. Statistical analysis and unsupervised learning have been applied in order to identify clusters of patients with respect to their mental state, quality of life and functional status at different time points.For each one of the above groups of variables, distinct clusters have been identified using the hierarchical clustering method. Regarding longitudinal clustering, the analyses conducted so far have led to the identification of sub-groups of patients with distinct trajectory patterns of psychological variables. Results have also suggested that specific types of psychological trajectories can be predicted fairly reliably based on data available at baseline.
The analysis and modelling outcomes will be further investigated using the data being currently generated by the running BOUNCE prospective pilot study. However, the specialized tools and workflows already created and the experience gathered so far justify encouraging expectations regarding a quantitative in-depth understanding of the complexities and the intricacies of life after breast cancer treatment, and subsequently the individualized optimization of the quality of life and resilience.
Full information on the project can be found here: https://www.bounce-project.eu/