Ģ��ֱ��

Ilkka Rautiainen works as a Doctoral Researcher in the Digital Health Intelligence Laboratory of the Faculty of Information Technology. In his dissertation, Rautiainen studied the potential of machine learning in predicting obesity and development of cardiorespiratory fitness. The dissertation also generated a personal health index, which is a potential tool for machine learning.

The first part of the dissertation consists of utilizing height and weight data collected from children at different times. The data was used in different machine learning models. The results were then compared with the reality: the number of children who had become overweight.

There were nine experimental designs in total. The predictions with a shortest time span were made for three-year-olds based on data collected up until they were two years old. Predictions with the longest time span were made for 13-16-year-olds based on data from their birth.

-The study confirmed that short-term obesity predictions are possible using machine learning, but longer-term predictions are much more challenging, says Rautiainen.

The second part of the study used a data-driven approach to predict the future development of cardiorespiratory fitness in young people. This was executed by using the “random forest” method. The method screens the most important factors from a large set of variables.

The study found that the development of fitness can be influenced not only by the baseline level of fitness but also by a wide range of factors, from physical well-being to psychological and social well-being.

-Poor cardiorespiratory fitness can make daily activities such as climbing stairs difficult. If fitness can be predicted reliably enough, it is possible to identify people with fitness challenges. Prevention could enable better help for these individuals, says Rautiainen.

According to Rautiainen, the findings of the study reinforce the importance of monitoring overall health beyond just physical well-being.

In the third part of the study, a method for calculating a personal health index was developed. The index aims to provide a holistic view of an individual's health and simultaneously considering the most common problems of health data. The health index is based on the ICF classification developed by the World Health Organisation (WHO).

-The health index provides a framework for combining data from different sources and has the potential to facilitate the use of data from a machine learning perspective, says Rautiainen.

Rautiainen emphasises that, as a whole, AI and machine learning methods enable better selection of individualised treatments, such as rehabilitation pathways. They also offer possibilities in developing treatments in a more preventive direction.

The dissertation of M.Sc. Ilkka Rautiainen is titled "Prediction methods for assessing the development of individual health status". The public examination takes place on Friday 1 March 2024 at 12.00 in Agora Auditorium 2 (Ag B105) and online.

Juha Röning from the University of Oulu and Kari Kalliokoski from the PET Centre, University of Turku will act as the opponents. The Custos is Research Coordinator, Adjunct professor Sami Äyrämö from the Ģ��ֱ��. 

The dissertation has been funded by the Jenny and Antti Wihuri Fund, David Health Solutions, the Faculty of Information Technology of the Ģ��ֱ��, and Business Finland.