Multiprofessional collaboration involving specialists from the Faculty of Medicine and the University's Global Campus team, which develops new methods of online learning in cooperation with faculties, led to the creation of a virtual contraceptive clinic.

According to University Lecturer Jaana-Maija Koivisto from the Faculty of Medicine, the idea for a virtual contraceptive clinic originated in a shared desire to develop a new kind of light and scalable virtual learning environment for teaching at the Faculty.

"A variety of virtual patients have long been used in teaching in the healthcare sector. However, introducing new options to teaching often turns out to be challenging: for example, the use of VR glasses is limited by their number and the effort required to create clinical content," Koivisto says.

Virtual encounters must be practised

The AI-based chat simulation was selected because it makes it possible for students to practise virtual patient encounters. Practising virtual appointments is important, as various chat services are becoming increasingly common in healthcare.

"While patients appreciate virtual services for the comfort and convenience they provide, healthcare staff perceive them as challenging: not seeing the patient makes the assessment of care needs more difficult. Training in written interaction must also include what are known as soft skills, such as expressing empathy," says Koivisto.

With suitable gynaecological expertise in the team, the conditions for developing a virtual contraceptive clinic were in place.

Artificial intelligence requires persuading

Through collaboration between the Faculty and the Global Campus project, four patient cases were created on the basis of the type of customers that usually reach out through chat services. Certain underlying factors affecting the choice of contraceptive were assigned to each AI patient for pedagogical purposes.

Saša Tkalcan, a specialist working in the Global Campus project, describes the creation of the AI patients as an iterative process.

"The AI patients were the result of collaboration, combining the knowledge of professionals from different fields. From our team Noha Abdelmonem and I provided the skills needed to formulate prompts for guiding the artificial intelligence, but experts are also needed to test the chat, to ask the right questions."

The process began with background interviews to train the chatbot to respond as a model patient, after which the prompts fed to the bot were modified on the basis of the observations made during the testing. Factors related to natural interaction emerged as a particular challenge.

"When we tested the first AI patient, a 17-year-old young woman from Helsinki, with experts, they let us know that the bot's manner of speaking needed further polishing. It was up to our team to think about how to persuade the artificial intelligence engine to function as desired," Tkalcan says.

Even though the model patients have been created, more work remains to be done on the virtual contraceptive clinic.

"We are already planning research on how interaction between students and the AI patients will turn out. In addition to the Faculty of Medicine, the aim is also to use the AI-based chat simulation this autumn with midwifery students from Metropolia University of Applied Sciences. This will provide us with new research knowledge on whether AI-based methods support learning," Koivisto points out.

According to her, the feedback from students so far has been encouraging: they have found the simulation setting realistic and considered the training meaningful.

The Global Campus project aims to develop more chat-based AI solutions for teaching.

"In the long run, our plan is to establish a model based on previous prompts that would enable anyone in the University community to create similar chat solutions for teaching. The development is at a promising stage," Tkalcan reveals.

Newsletter