Since joining the QoLEAD consortium as a PhD researcher, Baran Polat has been exploring how artificial intelligence can support people living with dementia at home. With a background in Natural Language Processing, a personal connection to dementia, and a passion for creating technology that people actually want to use, Baran combines technical expertise with a strong focus on real-world impact. We spoke with him about his research, motivation, and hopes for the future of dementia care.
From Enschede to AI research
Baran was born and raised in Enschede. He studied Artificial Intelligence in Nijmegen, completing both his bachelor’s and master’s degrees there before starting his PhD. His academic focus was Natural Language Processing (NLP), a field that studies how computers understand and generate human language.
Reflecting on the rapid developments in AI, he notes how quickly the field changed after the arrival of large language models such as ChatGPT.
“My focus was Natural Language Processing. Then suddenly ChatGPT arrived, and it felt like everything I had learned could be thrown out of the window. It became a completely new and interesting challenge.”
Today, he lives in Enschede again while continuing his research in dementia care and AI.
A personal motivation
For Baran, dementia research is more than an academic topic. Dementia has affected several people close to him.
His grandfather lived with Alzheimer’s disease and passed away more than ten years ago. Two of his uncles died from other forms of dementia at a relatively young age.
“My grandfather had Alzheimer’s and lived with it for a long time. Two of my uncles had different forms of dementia and passed away at around the age of 55. Seeing those situations gave me a lot of motivation for this PhD.”
His personal experiences helped spark an interest in developing technology that can make a meaningful difference in people’s daily lives.
Why QoLEAD?
After graduating, Baran explored several career options. He applied for positions both inside and outside academia, often focusing on topics related to ethics and healthcare.
At one point, he was close to accepting a consultancy role.
“I was about to sign a contract with a Big Four company. But as a consultant, you advise people without necessarily developing your own technical skills further.”
The QoLEAD position offered something different: the opportunity to continue programming, developing algorithms, and exploring new forms of AI while working on a socially relevant challenge.
“I wanted to continue developing hard skills such as programming and NLP. You can’t really advise people on these topics if you haven’t fully developed those skills yourself.”
Researching a virtual agent for daily support
Baran’s research focuses on developing a virtual agent for people with dementia who are living at home.
The virtual agent is designed as an AI character that can speak, move, and guide users through everyday activities. In the current project, cooking serves as the central use case.
“We are building a virtual agent that can help people with dementia who live at home. In our case, the focus is on cooking.”
The project is being developed in close collaboration with people living with dementia and healthcare professionals. Early this year, Baran and his colleagues organised focus groups to better understand what users actually need.
“We can design all sorts of things ourselves, but the first question should always be: what do people actually want?”
During these sessions, participants were shown a demonstration of the virtual agent and asked what they would like the technology to do in the future.
Adapting to different stages of dementia
One important challenge is that people living with dementia have very different needs depending on the stage of their condition.
Someone with mild dementia may require completely different support than someone in a more moderate phase of the disease. To address this, Baran’s research explores whether speech patterns can help identify these differences and enable more personalised support.
“Someone in a mild stage has different needs than someone in a moderate stage. The challenge is that the differences between those stages are often very subtle. Distinguishing between healthy cognition and Alzheimer’s disease is relatively straightforward, but recognising the differences between stages of dementia is much harder. To improve those models, you need more data and more information about a person’s condition. That often means relying on input from healthcare professionals, who are already under a lot of pressure and have very limited time. So, there is a constant balance between wanting enough data to improve the technology and not placing additional demands on the people working in care.”
For Baran, this illustrates one of the central challenges in developing AI for dementia care: creating systems that are accurate enough to respond to individual needs while working within the realities of everyday healthcare practice.
Balancing stakeholder needs
Developing AI for healthcare brings its own challenges. One of the biggest is ensuring that the technology serves the needs of many different stakeholders at the same time.
“Every stakeholder wants something different. A nurse wants one thing, an occupational therapist wants something else, and the person with dementia may have yet another preference.”
Co-design therefore plays an important role. At the same time, Baran recognises that involving people with dementia in early design stages can sometimes be difficult.
“At this stage of the design process, it can be challenging for people with dementia to form an opinion about something that does not yet really exist. It is often easier to evaluate technology after experiencing it in practice.”
Healthcare professionals, who often have more experience with existing technologies, are sometimes able to provide more concrete feedback in the earlier stages of development.
Beyond the PhD
Outside his research, Baran is also co-founder of a start-up that develops chatbot technology for people with dementia.
The platform focuses on cognitive stimulation and includes activities such as reminiscing with photographs, word games, music-based interactions, and conversations with a chatbot. The team is currently preparing for clinical trials after securing funding, creating exciting opportunities to further develop and test the technology in practice. At the same time, balancing a growing company alongside a PhD remains a significant challenge, as both demand substantial time, energy, and attention..
Tea, travel and kickboxing
Outside of his work in AI and dementia care, Baran has a passion that many people may not know about: tea.
Coming from a Turkish background, tea has always been an important part of daily life. Over time, his interest grew into a serious hobby.
“Some people have that fascination with brewing beer. For me, it’s tea.”
His enthusiasm has taken him around the world, including visits to tea regions in China, South Korea and his upcoming visit to Japan. One day, he hopes to open his own tea shop.
“PhD students often talk about opening up a bakery. Mine is to start a tea shop.”
Another hobby is kickboxing, which he started during his PhD. Inspired by his brother to give the sport a try, it quickly became an important outlet alongside the demands of research.
Looking ahead
When asked what success would look like in five years, Baran’s answer is clear: he wants people to genuinely use and benefit from the technology he helps create.
“A lot of healthcare technology ends up not being used because it doesn’t connect well with people’s needs. I want to prevent that.”
For him, practical impact matters more than personal recognition. Ultimately, he hopes the virtual agent will help people remain engaged in meaningful daily activities, whether that means continuing to cook independently or learning something entirely new.
A message for future researchers
Baran’s advice for future PhD students and professionals interested in dementia care is simple: leave the office and spend time with the people you are trying to help.
“Try to get out of your dusty office and connect directly with the people you are designing for.”
He has seen firsthand how valuable these interactions can be. During visits to De Meerpaal, one participant who was usually quiet suddenly started dancing when an Elvis Presley song played that an AI recommended.
“It was a welcome surprise that the tool could have such an impact.”
Moments like these remind him why user involvement matters so much.
“The more often you meet people and the longer you spend with them, the more open they become. That’s when you really start learning.”