The unique artificial intelligence experiment by the City of Espoo and the software and service company Tieto achieved the hoped-for result: artificial intelligence can pick up service paths out of an enormous mass of service data by grouping together risk factors that trigger the need for heavy and expensive services if found in the same person. The experiment was unique, because public administration client relationship data has never been combined and analysed as extensively using artificial intelligence before.
In the experiment that started in summer 2017, the social and health data and early education client relationship data of the entire population of Espoo for the years 2002–2016 were analysed by artificial intelligence. Data was collected on approximately 520,000 people and more than 37 million customer contacts.
Personal data was encrypted as early as the search phase, and all the data processing was done with extreme security. Individual persons could not be identified at any stage.
New information challenges the city
The experts of the City of Espoo were surprised by how extensive the data mass was and how well artificial intelligence could combine data. The experiment enabled, for the first time, the review of customer paths by family; the city’s data systems are usually individual-oriented.
– With the support of new technology, combining the data included in different data systems creates new data that can challenge the city. We have also received important information on the quality of the data and the registration practices, says Data Analyst Consultant Tomas Lehtinen.
Some observations can already be summed up. Artificial intelligence found approximately 280 factors that anticipated a child welfare client relationship. None of these factors alone is a risk, but the simultaneous occurrence of several factors can be a risk.
– The Espoo experiment demonstrates that artificial intelligence and the utilisation of data have a significant role in health care, whether in developing new service paths or supporting health care professionals in their daily work. The experiment also demonstrates that the quickly-implemented Intelligent Wellbeing platform by Tieto develops the data capability of social service and health care organisations, says Matti Ristimäki, Head of Data-Driven Businesses at Tieto.
Preliminary results on child welfare and child psychiatry client relationships
The data could be used to examine, for example, what differences there were in the previous use of services by families regarding which a child welfare report had been filed or that had received a referral to youth psychiatry.
The child welfare report or referral to youth psychiatry was preceded by using health services 3–5 times more than the control group. The family members in this group also used health services more than in the support group.
– The results, however, are very preliminary and they cannot as such be used for making conclusions. We need more research, experimentation and material analysis. The material did not include the use of private health care, and in the future, the data should be combined with the Kela database to specify the result, says Mari Ahlström, Director of Family and Social Services at the City of Espoo.
Development work is only beginning
The artificial intelligence experiment by Espoo and Tieto was unique because public administration client relationship data has never been combined and analysed as extensively using artificial intelligence before. The starting point was to test whether artificial intelligence can help target services proactively. The answer is a strong yes.
The data collected for the experiment and analysed by artificial intelligence has been destroyed.
– We now have a model that allows us to search for the equivalent fresh data whenever necessary. Therefore, the old data was not worth keeping, Tomas Lehtinen says.
Follow-up work will require, among other things, ethical consideration on whether the data produced by artificial intelligence could be used as an alarm so that the system would tell the care personnel whenever a customer arriving at an appointment has many risk factors. This is among the things that Espoo will discuss with experts of the field such as the Finnish Center for Artificial Intelligence (FCAI) founded by Aalto University and the University of Helsinki.
Besides FCAI, interest in artificial intelligence collaboration has been expressed, among others, by the National Institute for Health and Welfare and Kela.
– During the year, we have created a strong foundation for the future of municipal services. There is also a lot of interest within the city, and many ideas are under discussion, Päivi Sutinen, Service Development Director at the City of Espoo, rejoices.
– The data on Finnish health care is of a very high quality, and when it is collected in one place, such as a data pool, it can be better utilised, for example, with artificial intelligence for society, the development of municipal services, health care professionals and individual residents, Matti Ristimäki continues.
The experiment was one of Espoo’s contributions to the 6Aika project, in which the six largest cities in Finland develop services that are more open and intelligent based on the shared 6Aika strategy. Besides developing services, the aim of 6Aika is to create new competence, business and jobs in Finland.
6Aika in Espoo