Interview with WARIFA
We had the pleasure of speaking with two team members of the EU-funded WARIFA project, Conceição Granja and Thomas Schopf. Together, they answered our questions surrounding the aim, development and future plans of their risk assessment tool.
Conceição Granja is a Senior Researcher in Telemedicine at the Norwegian Centre for E-health Research (NSE), and Project Coordinator of WARIFA. She holds a PhD in Medical Informatics and conducted her postdoctoral studies at the NSE where she studied hospital-patient digital interaction in surgical contexts. Thomas Schopf practices as a doctor in the field of dermatology whilst also being employed as a researcher at the NSE. His main research interests are artificial intelligence for clinical decision support, teledermatology and web-based education on atopic eczema. At WARIFA, he is a researcher and member of the management team alongside Conceição.
What is the goal of the WARIFA project?
Conceição Granja: The goal of the WARIFA project is to use AI to provide our users with a combined risk assessment and recommendations that will help them improve the modifiable risk factors for the four most common chronic conditions. Combined means, when we make recommendations for one of the chronic conditions, we take the other three in consideration. For example, if we recommend type 1 diabetes patients to increase their physical activity by going for a walk, we could advise them not to do it during times with high UV levels, so their risk for melanoma won`t increase.
Which medical indications are you focusing on in the project?
Conceição Granja: WARIFA focuses on four chronic conditions: Both types of diabetes, type 1 in terms of supporting disease management and type 2 for prevention, cardiovascular disease, COPD and melanoma.
Thomas Schopf: I would like to add that this is only a proof-of-concept study, hence we are not expected to have a market ready app. Due to the short time frame, we need to focus on certain diseases and risk factors to show whether this concept can work.
“It is very important that a person with a disease does not forget to prevent other diseases.”
How do you define and assess patients' needs? Is there a standard that you apply or a specific structure?
Thomas Schopf: Yes, first of all you have to think about the two types of users. We have patients who already have a disease, but what is actually more important to us are the healthy citizens because this project is about disease prevention. But of course, we also want to emphasize the need if someone already has a condition. One of the underlying reasons for this project is that people are living longer and are getting more diseases at the same time. It is very important that a person with a disease does not forget to prevent other diseases. And this is exactly where we believe AI could make a contribution.
Conceição Granja: I can add a little more about our methodology. WARIFA is collecting three different types of data upon the user’s consent: Data from welfare devices, including smartwatches, smartphones, and medical devices, data from in-app questionnaires and so called technically ubiquitous data, for example the UV levels based on the location. It all started with an extensive literature review of the known and validated risk factors for each condition in order to develop the AI algorithms to predict the risk and now we are almost done with the Co-creation process, where we involve users in focus groups and interviews to get their feedback, not only on the looks and feel but on what they expect from the app. WARIFA also has user representatives as part of the consortium not only to help us reach out to people that have specific conditions, but also to inform us on the newest trends, wishes and expectations from users.
Do you have any current trends on the radar that you think are interesting or that you applied to your project?
Conceição Granja: On the technological side, the major trend is AI, not only to find patterns in the data but to enable personalized risk prediction. We all live in different places, have different diets and habits, therefore it´s very unlikely that two persons are exactly alike. With AI, we can find out which risk factors are relevant for a particular user and which factors have the greatest influence. WARIFA is not just collecting diabetes related data to make an assessment for diabetes. WARIFA is collecting data for all four conditions to make a risk prediction. We might even see that there are risk factors that are not specifically or directly related to diabetes but are influencing that condition nonetheless. With the help of AI, we can give users a more holistic view on what is actually affecting their health.
Thomas Schopf: In the medical community, many physicians used to be very skeptical about AI because they thought it might replace them. However, I think that it has now been recognized that these instruments aren´t supposed to replace but assist the doctor. It is a shared decision-making process in which the doctor has the final say, but with the help of AI, decisions will be made easier.
“With the help of AI, we can give users a more holistic view on what is actually affecting their health.”
Which role does mental health play in your disease prevention?
Conceição Granja: Due to the time frame of our project, mental health is currently not a focus topic for WARIFA. However, we are considering how we can reach people and we know that in the future this will imply the inclusion of mental health. At the moment, we are not asking directly about mental health, but we are trying to adapt the way we phrase a recommendation as to not overwhelm people or make them afraid because they may be at high risk for a particular disease.
How do you plan to integrate WARIFA into the current clinical practice and which implications will WARIFA have on the healthcare systems?
Thomas Schopf: WARIFA is primarily intended for healthy home-based users. But we are very interested to see how this could fit into the healthcare system. Therefore, part of our proposal is to look at new possible healthcare pathways. The first step to that would be a stakeholder analysis. We have already conducted Interviews with doctors and key stakeholders, asking them about their thoughts and ideas. And I think the next step would be to find a way to transfer the data WARIFA calculates to healthcare records. But based on my experience with the healthcare system, I know it will take a lot of work to make the systems compatible.
Conceição Granja: I wonder what impact our collected data and risk prediction could have on general practitioner appointments. A user may not need to visit their GP for a risk assessment because our app can carry out such an assessment. If the calculated risk is high, then WARIFA would recommend the user to consult their GP. The number of first appointments for risk assessments could drop significantly as the app cannot only provide a quick assessment but also recommendations. WARIFA can also be helpful during GP appointments, as patients can provide a wide range of health-related data, their diet, physical activity habits, smoking, alcohol consumption, and so on. The GP will be able to make a quick and informed decision which will contribute to the efficiency of healthcare systems. We could therefore have fewer and shorter appointments.
Which tools do you give users to sustainably change their lifestyle, and how do you implement the lifestyle changes into the app?
Conceição Granja: WARIFA's entire focus is on the sustainable change of lifestyle habits. We are approaching this by providing users with recommendations and additional literacy information. Because some people like to read scientific papers explaining why a certain lifestyle puts them at risk, while others prefer shorter information, we personalize the type of information. As with the type of information, some people respond better when we point out what is wrong with their actions, while others like to know what they are doing right so they can keep going. See for example the difference between a recommendation for a professional athlete compared to someone who spends most of their free time on the couch. The recommendation to walk to the supermarket instead of taking the car may have a much stronger effect on the second person, because for them, even a 15 minute walk a day would be a good achievement and should be communicated through positive reinforcement. By customizing the type and frequency of communication with our users, we hope to keep them engaged with the app. We would also like to use some kind of gamification but are a bit limited due to the time frame of the project so we´ll see to what level we can introduce it.
How do you want to evaluate the effectiveness and safety of your app?
Conceição Granja: WARIFA is a proof of concept. The aim of the project is to prove that we can use AI in a meaningful way for risk prediction and lifestyle recommendations. We will run a pilot study in which the data we collect from users and the recommendations provided by the algorithms will be manually evaluated by doctors. We have doctors who are experts in all four chronic diseases involved in the project and they will validate the accuracy of the algorithms. Safety wise, we can look at data security and medical safety. In terms of data security, everything we do on WARIFA, from data collection process to rights we give to our users, is being validated by a data protection officer. The medical safety is assured because WARIFA is mainly working with prevention and if we see that a user is at high risk, we will ask the person to consult with a medical professional. The management of type 1 diabetes is also part of WARIFA's scope. We are providing users with predictions of their glucose levels, but we are not recommending changes in their treatment plan. If we determine from their current data that they need an adjustment, we will recommend that they consult a doctor.
WARIFA has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 101017385.
This output reflects the views of the authors, and the European Commission is not responsible for any use that may be made of the information contained therein.