Every year, around 300 children are diagnosed with cancer in Sweden. Opportunities for treatment are good, and four in five survive the cancer. But many who have survived cancer as a child have different problems later in life. Scientists at RISE will now use applied AI to investigate whether it is possible to find patterns among those affected.
Children who have survived cancer are often affected by so-called late effects in adulthood. Late effects constitute long-term changes that persist after treatment is completed. This may include reduced growth in height, heart failure, infertility/sterility, hearing or vision impairment, or learning disabilities, but so far there has been no clear picture of who is affected.
“With the help of artificial intelligence and machine learning, we hope to identify groups that are at risk of late effects,” says Anders Holst, researcher at RISE.
Anders Holst and his research team are now looking at data from all childhood cancer survivors from the 1970s – around 2,000 individuals. The amount of data makes it possible to find correlations between different patients’ disease histories.
“We will look at the correlation between personal information, diagnosis, treatment and late effects,” says Holst. “To do that, we cannot look at individuals one by one, but we divide different cases into clusters.”
Correlation between diagnosis, treatment and late effects
The researchers are also investigating the correlation between cancer diagnosis and late effects versus treatment and late effects. In other words, is it the cancer diagnosis that causes complications later in life or is it the treatment? According to Anders Holst, we don’t know for sure:
“We must separate direct causes from indirect to find the answers.
The results are expected to create new hypotheses regarding which treatments are causing or contributing to complications, which in turn may present new preventive measures, better treatment strategies or the development of new low-risk therapies.
The project also aims to present personalised follow-up advice for each individual childhood cancer survivor’s therapy data. By tailoring how therapy data is presented to each individual, the potential of obtaining information related to their own health increases. In addition, presenting the information in a user-friendly and secure manner simplifies the work for hospital staff.
The project has numerous objectives: information security, personalised care, the right care at the right time, prevention and the possibility of identifying new risk groups. However, an overall goal is to help reduce human suffering – the best possible use of artificial intelligence and machine learning.
The project started in November 2017 and will run until May 2020. In addition to RISE, the Department of Clinical Sciences, Lund University, Climber AB and Region Skåne Paediatric Oncology are also involved in the project, which is being funded by Vinnova under the title Artificial Intelligence for Better Health.