The interdisciplinary team behind Adhermo combined medicine, data science and software development to explore whether patient adherence in type 2 diabetes could be predicted and supported through a single application.
For people living with type 2 diabetes, following prescribed treatment consistently is central to managing the disease. For family doctors, however, monitoring adherence across a large number of patients is difficult, particularly when the signs of emerging problems are spread across different records and patterns of behaviour.
A team of students from the University of Zagreb School of Medicine, the Faculty of Science and the Faculty of Electrical Engineering and Computing developed Adhermo as a possible response to that problem.
The application integrates a predictive model designed to estimate adherence among patients with type 2 diabetes. It is intended to help family doctors monitor patients’ status, recognise potentially risky patterns and receive guidance relevant to treatment follow-up.
The team also designed Adhermo to provide support to patients while they are following their therapy.
Its aim is therefore broader than producing a prediction alone. The application is conceived as a tool that could help doctors and patients follow treatment more consistently and effectively.
Three disciplines, one clinical problem
The composition of the team reflected the different kinds of knowledge required to develop the prototype.
Medical students contributed their understanding of type 2 diabetes and the clinical problem of treatment adherence. Students from the Faculty of Science and the Faculty of Electrical Engineering and Computing brought experience relevant to working with data, predictive modelling and application development.
During the competition, the team expanded its knowledge of large databases, model development, optimisation and model analysis. It also studied the medical literature on type 2 diabetes, improving its understanding of the disease while developing the technical solution.
This combination was important because the project required the team to move between two different questions: what information could be extracted from the data, and how that information might be used in the context of diabetes care.
Learning the full development process
Before taking part in the AI4Health.Cro Innovation Challenge, the students had not had an opportunity to work through every stage of developing such a solution.
The competition allowed them to follow the process from an initial idea to a predictive model and then to the development of an application prototype. For the team, this insight into the complete development process was one of the most valuable parts of the experience.
They began with a defined healthcare problem, worked with the available data and used the results to develop the model on which Adhermo was based. The process also required them to examine and improve the model rather than treating its first output as a finished result.
In doing so, they developed practical experience in areas that had previously been encountered primarily through their separate academic disciplines.
What AI4Health.Cro provided
The AI4Health.Cro competition gave the team access to the volume of data needed to develop both the predictive model and the application prototype. The students identified the clearly defined problem and the availability of a large dataset as two of the most useful elements of the process.
Equally important was the support provided by the organisers. When the team encountered an unclear issue or needed guidance, it could contact the AI4Health.Cro team. According to the students, this meant they were not left to resolve every difficulty on their own.
They described the support as continuous throughout the competition, from the early stages of working with the data to the development of the final prototype.
For a student team working across several disciplines, this structure provided a way to connect knowledge acquired through formal education with the practical demands of developing a digital health solution.
The competition did not simply give them a problem to solve. It also gave them access to the data and expert support required to move from a concept towards a working model and application.
A prototype that still needs direction
The team is interested in continuing the development of Adhermo. Its members would like to move beyond the current prototype and build a solution that doctors and patients could use in everyday practice.
They are also clear about what they do not yet know. The team has limited experience of the relevant market and of the most appropriate route for developing the prototype further. Its next step would therefore include consultation with an expert who could help assess the available options and advise on the direction of development.
This may involve determining how the application should be refined, what would be required for further validation and how a student prototype could be developed into a product intended for practical use.
The team plans to use part of the competition prize for further education and part for travel. It also intends to use the opportunity to explore what more could be built from the existing prototype.
Adhermo remains at an early stage. The information provided by the team does not yet establish how the model would perform in clinical practice or how the application would be introduced into routine care.
What the competition provided was an initial route through the development process: a defined medical problem, a substantial dataset, support when technical or methodological questions arose and an opportunity to connect a predictive model with a usable application concept.
For the Adhermo team, that process transformed separate areas of student knowledge into a shared healthtech project and gave its members their first experience of taking a solution from an idea, through model development, to a working prototype.