Assistant Professor Pekka Marttinen develops machine learning solutions to real-life problems

What type of research do you do?

I study and develop machine-learning methods related to medicine and biology. I cooperate extensively with different partners, but at the same time, my method development work is independent.

Together with the National Institute for Health and Welfare (THL), I develop analytical methods for predicting different population groups’ long-term health. These statistical forecasting models can support the decision-making of the National Institute for Health and Welfare. The models can be useful when defining the capitation payments of the health and social services reform, that is, the amount of state compensation provided to suppliers of different health care services. The purpose of the model is to provide different population groups with the most appropriate resources.

The capitation fees are just one example of the interesting scientific problems that the new machine learning methods may help with. I aim for the most accurate forecasts, and I am also involved in the expert groups of the health and social services reform.

The predictive models I have created can also be used to identify risk groups in the population. They can be useful when assessing the need for screening the whole population or a large part of it, or monitoring the health of pregnant women suffering from gestational diabetes. The models may also provide individual and supportive recommendations on how to maintain optimal glucose levels.

I also make predictive models for plant breeding. They make use of data such as satellite images, weather, and individual properties of fields. The effects of hot summer seasons on yield and their potential impact on future cultivation is also a crucial topic to consider.

How did you become a researcher?

In 2003, I had a summer job at the Rolf Nevanlinna Institute of the University of Helsinki and worked in a biometry research group, doing research in applied mathematics and developing statistical methods for the problems in biology research.

I did my master's degree in maths and my dissertation in statistics, both at the University of Helsinki. Further studies in the Centre of Excellence in Biometry allowed me to continue with the topic of my master’s thesis.

After my dissertation, I managed to work for half a year at a consulting company when I heard about the position of a teaching researcher at Aalto University. A career as a researcher seemed interesting, fun and important—you get to educate future generations.

What are the highlights of your career so far?

There have been many highlights, often connected with the completion of something. After the last lecture of a course, I often take the rest of the day off to have time for myself. A similar special moment in the work of a researcher takes place when an article is sent for review or approved.

I spent the academic year 2013–2014 at Harvard, Boston, with my family. There I was able to focus on research, together with engaging topics and cooperation partners. It was a successful experience and brought us closer together as a family, too.

I also really enjoy communicating and brainstorming with students in peace. It is interesting to cooperate with different people from diverse backgrounds.

Every day can have special moments. I might listen to some music at work or write a diary in the evening before going to sleep.

What is the most essential characteristic of a researcher?

The answer depends on the field of research. My research is interdisciplinary and requires patience, perseverance and an ability to communicate with experts from different fields. A project may take longer than anyone would ever have guessed, but finally everything gets done.

What do you expect from the future?

I do my best at work, one day at a time. On the other hand, public examinations of my own students’ doctoral dissertations will certainly be special moments.

Further information:

Pekka Marttinen
Assistant Professor
Aalto University, Department of Computer Science
tel. +358 44 303 0349
[email protected]