Researcher profiles

International project on social robots and child mental health

One moment, Ginevra Castellano, Professor of Intelligent Interactive Systems and Director of the Uppsala Social Robotics Lab, who is coordinating a new international project, ‘Measuring children's well-being and mental health with social robots’ (MICRO).

The project has received just over SEK 17 million in the CHANSE-NORFACE Enhancing Well-being for the Future call, with Uppsala University's part funded with just over SEK 5.7 million by the Swedish Research Council and Forte.

What is this project about and what do you want to achieve?

- The aim of the project is to increase knowledge about how language abilities and the ability to successfully participate in social interactions affect children's well-being and mental health. To achieve this, we will explore the use of social robots as new tools to measure children's wellbeing and mental health in an educational setting. This will be done with a particular focus on vulnerable groups that are potential targets for preventive interventions, such as children with language development disorder and refugee children.

How is it possible to measure children's well-being and mental health with social robots - can robots really do it?

- We will use already existing validated well-being and mental health questionnaires that will be administered by a social robot. Then we will also develop new methods to measure well-being by analysing behaviours that occur in interactions between children and robots.

Machines that reason about what they see

Nataša is a professor of computerised image analysis. She develops algorithms that enable computers to efficiently and reliably extract and interpret information from digital images.

- We humans process enormous amounts of information through our visual system and consider sight to be the most important of our senses. Developing methods that enable the automated analysis of the vast and rapidly growing amount of visual data collected by a multitude of different sensors in various application areas is not only essential for the development of these fields, but it also interests people on a fundamental level. We simply like to teach our computers to “see” what we can see ourselves!"

Nataša began her academic career in mathematics and eventually developed an interest in visual data analysis. “My background is in mathematics, and my entry into the field of digital image analysis was through discrete geometry, which is one of the fundamental theoretical frameworks for representing and interpreting digital images. I became fascinated by the intuitiveness and interpretability of many theoretical results in discrete geometry when they were put into the context of visual data analysis.”

In this field, it is not uncommon to collaborate with researchers and professionals from many different parts of society. This is not surprising, considering how important visual data processing has become and how versatile this data is.

“Computerised image analysis is a highly interdisciplinary field. Images are signals, big data, abstract mathematical structures, and visual representations of phenomena in a multitude of applications. It usually requires the collaboration of a whole range of experts – physicists and engineers, computer scientists, statisticians and mathematicians, and specialists with domain-specific knowledge – doctors, astronomers, archaeologists, life scientists… – to enable a computer to do what we humans can do so naturally – process and interpret visual data.”

“In the life sciences, powerful and complex imaging techniques can reveal a multitude of properties of an object of interest – morphology, dynamics, function – but usually only one such property at a time,” explains Nataša.

To achieve a comprehensive perspective, we must combine several different techniques, including AI-based ones. “For a complete understanding of the objects of interest and the processes involving them, various complementary techniques must be combined. I am interested in the development of AI-based methods that enable reliable, interpretable, and dependable use and analysis of information-rich multimodal image data, particularly in biomedicine.”

“AI-based methods, and particularly those built on deep learning, are now dominant in the field of image analysis due to their generally outstanding performance. They have the potential to enable the integration and analysis of heterogeneous multimodal information in the most challenging scenarios,” she says.

“A prerequisite for the utilisation of AI-supported analysis and decision systems in biomedicine and healthcare is the interpretability of the obtained results. Doctors must be provided with ways to understand the basis of the decisions, allowing them to determine whether they can trust the system. We strive for solutions based on Explainable Artificial Intelligence (XAI). Reliable AI systems will open many pathways to the discovery of new knowledge in science by highlighting relevant patterns and relationships in data that we were previously unaware of.”

The development of these methods is an active and growing research field; Nataša’s research group, MIDA (Methods for Image Data Analysis), has proposed several new ways in which AI can be used for improved image analysis.

“We are developing AI-based methods for multimodal image analysis, particularly in the context of cancer detection. Our methods for multimodal image registration achieve state-of-the-art performance in several multimodal biomedical and medical application scenarios. The methods are shared publicly and used by other researchers. We are now integrating them into our developed AI-based decision support system for the detection of oral cancer. We believe that by using versatile multimodal information, we can increase both the reliability and interpretability of the system.”

The MIDA research group actively collaborates with experts in life sciences and healthcare. “Several former and current members of MIDA, the research group I lead in the IT department, have close collaborations with colleagues from life sciences and healthcare. Together, we plan to apply our developed methods in various biomedical scenarios and evaluate their potential to enhance everyday healthcare, particularly for improved cancer detection, but also for a better understanding of the disease.”

Nataša is also a group leader in the international network COMULIS. “A large European COST network, COMULIS (Correlative Multimodal Imaging in Life Sciences), where I serve as a working group leader, provides a rich international interdisciplinary scientific environment that further stimulates synergies towards exciting research in life sciences and the development of advanced computational techniques.”

But it also comes with its own challenges. “As is often the case in highly interdisciplinary fields, a major challenge is learning to communicate with colleagues from different scientific backgrounds, finding a common language across disciplines, and understanding the needs and limitations of the various methods involved. However, this is crucial for success, and once achieved, it is also very inspiring and rewarding.”

Some concluding words from Nataša. “Digital images are always just approximations of objects and scenes from the real world. Imaging techniques are limited by the underlying physical processes, as well as optimised imaging conditions (e.g., to prevent potential damage). Even though the amount of collected data is enormous, its informative content often seems insufficient. Researchers (and people in general) will always want to see more, smaller, faster events and objects – to imagine the world with more details and sharper colours.”

“I have worked extensively on developing methods that maximise the information extracted from available data. These include methods to compensate for limited image resolution when performing precise measurements of objects, approaches to enhance the quality of images taken under specific constraints, and – what I am most interested in right now – methods to meaningfully combine heterogeneous information from images acquired by different sensors to achieve new and higher levels of understanding of complex phenomena,” says Nataša.

And in the future?
“Considering that modern image analysis heavily relies on artificial intelligence, machine learning, and data-driven approaches, a very important next step is to develop methods to teach our automated systems to ‘reason’. To go beyond detecting patterns (correlations) in data, towards understanding causality and (perhaps even) achieving truly intelligent behaviours.”

“However, this is a long-term goal, but we are continuously making small steps forward. We are very interested in being part of that journey!”

Interviewed by Victor Kuismin, 18 March, 2022

"Calculations that Save Resources and Lives"

Elisabeth Larsson’s research focuses on computational science. By creating numerical models and simulations of how muscles, airplanes, or the financial market work, both resources and lives can be saved.

- Computational science is the best subject in the world because we have the freedom to do almost anything we want! says Elisabeth, a senior lecturer in computational science at the Department of Information Technology at Uppsala University. It is an incredibly interdisciplinary subject. We border on mathematics, computer science, and all conceivable application areas – it can be anything from humanities to technology.

One interdisciplinary project that Elisabeth has been working on since 2012 is a collaboration with Dr. Nicola Cacciani at Karolinska Institutet and Pierre Villard, a French image analyst and biomechanic. What we are trying to do is build a simulation model for the respiratory muscles, a computer model of the human respiratory system.

From a medical perspective, the simulation model is intended to study how a respirator, or ventilator as it is actually called, affects patients (ventilator-induced diaphragmatic dysfunction).
- Normally, when you breathe, the diaphragm contracts so that the lungs expand and air is drawn into the lungs, but when a patient is placed in a ventilator, the air is instead pushed into the lungs, explains Elisabeth. This puts a strain on the diaphragm muscle, which impairs its function.

- This happens quite quickly,” says Elisabeth. In 24 hours in a ventilator, you lose 30 percent of the muscle function in the respiratory muscles. This means that a patient becomes both healthier and sicker during intensive care, and it takes time to rehabilitate this damage.

The idea of conducting experiments on the respiratory muscles using computer simulations instead of animals is to avoid harming anyone.
- The doctor is interested in the medical question, but I am interested in the numerics, says Elisabeth.

A result of the research so far is shown by Elisabeth as an image of the diaphragm in the respiratory muscles.

-We need to find a way to describe the muscle to the computer using numerical methods,” says Elisabeth. What we do then is pick out a number of points and describe how the points relate to each other mathematically. It is a numerical model.
- But having a model is not enough – then I have to do a numerical simulation of the breathing process. We need to decide what the computer should do with the model. The description needs to come to life in the computer. Then we write algorithms that work with the model. We need to come up with the algorithms mathematically and implement them, which is what we call it when we write a program that describes the algorithms in a language the computer understands.

To be able to create the numerical models of the muscle, Elisabeth herself has to understand how the muscles work.
- The core of our research is to know and understand when a model can be trusted, when it will give a correct answer. Computational science is important because simulations are used in so many areas, such as when building airplanes or bridges. If you do not understand the limitations of the simulation, it is a dangerous tool to use.

Elisabeth’s interest in mathematics was sparked early on.
- I was interested in counting even before I started school, says Elisabeth. It was obvious that I wanted to study mathematics because it was my great interest. But it was only when I started studying computational science that I realized – yes – this is what I love! To be able to calculate and get an answer.

Today, the focus of Elisabeth’s research is on method development.
- Then the method development leads to working with very exciting applications. I think it is fun to solve real problems because it is much more difficult.

The research project on the simulation of the respiratory system started when they were contacted by Nicola.
- I didn’t think we could say no to a doctor who comes to us and wants to do something mathematical, says Elisabeth. But no one else wanted to take the project because it was so complicated, so I took it.

Elisabeth’s research contributes, for example, to solving medical problems, but she has also worked on solving financial problems where models are developed to create increased stability in the financial market through more accurate pricing and risk calculations.

Computational science has great relevance for adjacent research areas. Computation is used in engineering and natural sciences. It is also used in the humanities and social sciences, but it is still on the rise there.
- What I love about computational science is all these collaborations in different directions because it means I get to learn about different fields – statistics, medicine, or finance. I have to learn the field to have a connection to what is relevant.

The biggest challenge is funding the research. Elisabeth believes that one problem is that the major results from the research are “invisible” when they are within the scientific community.

- It is difficult to assert oneself when you are a subject seen as a support for other sciences and to get funders to appreciate that what we do has value in itself, says Elisabeth. But I actually think things are going well right now, it’s just a matter of finding the time to do everything well.

Elisabeth’s immediate goal is to organize the researchers around the world who work with radial basis functions and ensure they collaborate.

- We are going to write a common code that everyone can use, says Elisabeth. We will have our first meeting in September in Italy.

- I think computational science is so incredibly cool, so it feels like everyone else should think so too. The fact that you can input a problem into the computer and gain knowledge about the problem is amazing. Using computer simulations can save both resources and lives.

”Good democracy needs good communication”

How can we increase understanding of how our communication and information dissemination online and on social media affect societal development? Matteo Magnani and the research group Infolab collaborate with sociologists and study, among other things, how information arises and spreads between different platforms on the internet.

Matteo Magnani’s research focuses on developing methods to analyse online data, particularly user-generated content on social media, to gain a better understanding of society.
- Here at Infolab, we work on the technical aspects of analysis methods. We collaborate with sociologists who help with data understanding and analysis, says Matteo Magnani, an Associate Professor in Computer Science at the Department of Information Technology at Uppsala University.

Examples of analyses can include identifying key individuals, those driving discussions on certain topics in the political debate on platforms like Twitter. Or identifying online conversations – being able to gather short comments or small pieces of information that pertain to the same topic.
- We work extensively in the field of “social network analysis”. In this area, simple graphs have been used for many years. With graphs, social networks are modelled with circles representing, for example, people and lines representing the connections between these people, explains Matteo. However, this model is not sufficient for analysing online data.

There are different types of interactions within a platform. On Twitter, for example, you can reply to a tweet or retweet and comment. These different interactions have different meanings, and mathematical models are needed to distinguish between them. This is what the Infolab group is researching.
- We are also interested in how information, such as fake news, spreads online. This happens through different layers of interaction on various platforms.

According to Matteo, perhaps the most important result of their research is that they have developed methods for analysing so-called “multilayer networks”. Matteo is one of the authors of the book Multilayer Social Networks. They have also created the program multinet, which is a tool for analysing multilayer networks.
- In recent years, I have travelled to various conferences to talk about multilayer networks, our methods, and software for analysis, says Matteo. Among others, at Sunbelt, the largest social network analysis conference in the world.

Matteo explains that his interest in researching this interdisciplinary field is largely due to his connection with a childhood friend from Italy.
- Luca, who now researches at the IT University of Copenhagen, studied sociology while I studied computer science. Ten years ago, we started discussing how we could use online data in social research. That’s how it began, and we still collaborate. He was one of the co-authors of Multilayer Social Networks.

Matteo also felt that information technology, especially in his field of databases, did not focus on people but on technology.
- Information technology is so important for us as individuals – we receive information through information technology. We communicate, we create our ideas, and we change our ideas on social media, among other things.

Matteo wants to be part of solving the major problems in society. One of these is the political development with the spread of populism. He points out that it is not information technology that has caused this development, but information technology still plays a role, for example, through the dissemination of information and fake news.
- So for me, it is important to find out how we can use our knowledge in computer science to do something good for society.

Matteo and his colleagues have just received funding from NOS-HS (Nordic Research Council for Humanities and Social Sciences) to build a Nordic network on online disinformation and its impact on society.
- We will organise three Nordic workshops, says Matteo. For this network to be effective, collaboration between different fields is necessary; it cannot be just computer science or sociology. Analysing WHAT is disinformation is complicated. Who has the right to consider something true or false? In political texts, facts, values, and emotions are mixed.

- Complexity is a key word for me – reality is complicated, and we need to find methods to address this complexity when analysing online data. You cannot create good solutions with just a mathematical abstraction, an algorithm, or a social theory that does not look at the existing data and show what is actually happening. We need to collaborate for it to be effective.

Currently, the research focuses on text, time, and uncertainty. Colleague Davide Vega D’aurelio is working on methods to include text analysis, and Amin Kaveh is studying how to analyse networks when only the probability of connections between people or text is known. The plans are to find funding to further develop this and establish collaborations with linguists.
- It is also important in the analysis to know WHEN things have happened online, says Matteo. Together with my colleague Christian Rhoner, we are collaborating with two researchers from Tokyo Tech to develop analysis methods for when interactions occur. This will help us not only understand WHO is talking about WHAT but also how discussions develop and change over time.

Current research focuses on public figures and their interactions that are obviously made public on social media. GDPR and the protection of private individuals online make it difficult to research social networks there. Obtaining consent from private individuals can be challenging, even impossible, when platforms restrict how their members can be contacted. Matteo and his colleagues have written an article that problematises this and provides a practical guide together with sociologists and a lawyer.
- This ethical and legal discussion is important, what are we allowed to do? says Matteo. We, as researchers in computer science, need to think more about this. Information must be free, but we must also be able to protect individuals.

Currently, Matteo and his colleagues at Infolab are working to secure more resources to expand their research and further develop methods and models for network analysis.
- We need more complex models that can take more factors into account because we want to study society as a complex system, says Matteo. A good democracy needs good communication – this is what we aim to contribute to with our research – an increased understanding of how communication is used online and affects society.

In addition to his research and role as Director of Studies, Matteo is also responsible for developing a new Data Science Arena, which will be a network for everyone working in or interested in data science. He is also leading the creation of a new international program in Data Science. In the time that remains, Matteo is also a teacher.
- It is a super important role for me! says Matteo. There, I feel that I am making a contribution to society. By teaching people who will be significant for society, we teachers play a big role. Besides teaching technical skills, I also want to teach the ability to think about consequences and ethics. To create an understanding that the solutions we develop in information technology can be quite problematic. There, I feel that what I do is really important!

“We make the invisible visible”

In the future, maxillofacial surgeons will be able to perform the operation on a computer before they actually open the patient. Using scanned three-dimensional images of the injured patients, computer models are created that give the surgeon better control over the actual appearance of the injury. The invisible beneath the skin is made visible. “The research will result in shorter operation times, better surgical outcomes, and saved money,” says Ingela Nyström, Professor of Visualization at the Department of Information Technology at Uppsala University.

Ingela Nyström’s research focuses on the intersection of mathematics, information technology, and medicine. She is a mathematician and computer scientist by training, but already as a doctoral student, she discovered that her applied mathematics could be useful in medicine. The research involved developing methods for, among other things, examining shapes in images on the computer.
- When my supervisors and I inquired about the interest in this, it turned out that the radiologists at the hospital were interested in being able to examine blood vessels on the computer. We then worked on mathematically examining where there were blood vessel constrictions.

Ten years ago, they made contact with maxillofacial surgeons at Uppsala University Hospital, and the collaboration took off. The surgeons needed more precise methods to plan jaw surgery.
- With this methodology – if someone has been injured in a fall or kicked by a horse – a CT scan of the injured jaw can be made, and then the broken jaw can be assembled like a 3D puzzle on the computer before performing the actual surgery, explains Ingela.
- What we have built into the system is advanced image analysis and visualization so that one can see how deep an injury is or how shattered a bone fragment actually is before opening up the patient during surgery. You could say that we make the invisible visible!

Working with mathematics, data, and medicine was something that felt natural to Ingela from an early stage.
- I have always wanted to work with people; for me, a career in healthcare could have been an option, says Ingela. But if I had not studied mathematics, I would have missed it every day; I am probably a scientist to my core. However, I am not a mathematician who just wants to delve into formulas; there has to be this element of usefulness.

Ingela’s grandmother was a bit worried when she started studying mathematics, but at her dissertation, she expressed that mathematics wasn’t so bad after all, because with it, Ingela could help in medicine.
- It will go well for you, Ingela, my grandmother said to me then.

The current focus is on continuing the collaboration with the maxillofacial surgeon.
- We have solved the puzzle pieces with bones – assembling and making implants. But now we also want to look at the soft tissues, as it is also necessary to be able to transplant skin or move muscles, and that area is very “under-researched”. We have just started with basic research in this area, but have not yet found the right funding for the research.

The latest results of the research are that they can draw on the skin using patient-specific data. They can look at the skin flap that needs to be removed and then find the corresponding place on the skin on the inside of the thigh. With the help of the program and scanning from the patient, they can find healthy tissue in a place where there is also a good blood vessel. By attaching the skin flap with a good blood vessel to the face, it heals much faster.
- This method is something that plastic surgeons have started to look at, says Ingela. A real challenge in the research is to create models that simulate soft tissue, tendons, and muscle tissue. It is more complex to model flexible, soft materials than the rigid skeleton.

Something that runs like a common thread through all of Ingela’s research is precision.
- You have to be mathematically correct, otherwise, the results will not be good, she explains. But also precision in ensuring that patient data is handled correctly, and precision in making the programs user-friendly so that they work for the surgeons who will use them.

The methods that Ingela and her colleagues develop will be able to contribute to shorter operation times and better surgical outcomes, which benefits the patients. As a result, these methods can also help save money in healthcare.

The method with the computerized 3D puzzle could also be useful in other fields, such as archaeologists who need to piece together ancient urns using fragile shards of pottery.

Ingela now wants to grow the research group. There are several surgeons and doctoral students at Uppsala University Hospital who are active in the project, but at the Department of Information Technology, the group is small.
- Today, we are Fredrik Nysjö, who is a doctoral student, Filip Malmberg, who is a researcher, and myself, says Ingela. So we need more people in IT.

Until now, Ingela has also been the head of the Vi2 division, the Division of Visual Information and Interaction. Besides getting more time for her research now that she is handing over the role, she looks forward to teaching again.

- Being a teacher is the most fascinating thing one can be! exclaims Ingela. To pass on what I have learned to the students.

If a student is interested in researching in the same field as Ingela, she advises them to first focus on learning mathematics and then become a strong programmer. After that, they can start building bridges to medicine.

”I am somewhat of an entrepreneur”

Thomas Schön has built a successful research team in Machine Learning at the Department of Information Technology. There is a buzz of activity in many different research areas, largely thanks to Thomas’s ability to forge collaborations and drive projects forward.

- My research focuses on Machine Learning, which is a part of AI, artificial intelligence, says Thomas Schön, Professor of Automatic Control at the Department of Information Technology at Uppsala University. It involves building mathematical models from data and enabling computers to learn things they are not specifically programmed for.

Thomas’s interest in Machine Learning was sparked a few years after his PhD for two reasons.
-Firstly, it uses the mathematics that I know and enjoy, and secondly, it applies this mathematics to a much broader set of questions than I was used to, giving me access to many new and exciting applications. I also believed that Machine Learning would have a significant impact, but the main reason was probably that I simply couldn’t resist getting into the field.

There are many different applications for Machine Learning, and there is a lot happening within Thomas’s research team. In fact, it seems to be buzzing with activity.
- One of the projects we are currently working on is the automatic interpretation of ECGs. This is a project where we collaborate with researchers in Brazil. Our automatic interpretation is actually better than real doctors at detecting five of the six most common heart conditions now. These are brand new results that we have just published!

Thomas explains that he strives to deliver the best fundamental research and the most relevant applied research. The applied research is conducted in collaboration with companies or applied research groups.
- For example, the research on the automatic interpretation of ECGs, where cardiologists in Brazil have been collecting data for ten years and have created a research group, Thomas explains. A visiting PhD student at the department comes from that group. He presented the project and asked if I wanted to help, and it was a perfect fit. We helped with the mathematical models (deep learning in this case) and they contribute with the clinical knowledge, resulting in a good use of our skills in a joint application.

The foundation is that Thomas’s research team further develops mathematical models and examines their properties and how they can be used. The specific application they research is often random. As long as the research can benefit society.
- A collaboration can arise from a meeting with someone to discuss something completely different. In fact, I expose myself to situations and meetings to see what ideas might emerge.

Ett av fokusområdena för Thomas forskning är medicin.
- Det startade med att jag fick bra kontakter med läkare bland annat genom en ledarskapskurs som jag gick. Uppsala är bra på det sättet genom att vi finns nära Universitetssjukhuset och läkare som forskar. När sedan min mamma blev sjuk i cancer påverkade det mig naturligtvis personligen, så jag ville hitta något sätt som jag kunde bidra till att göra den vården bättre.
När företaget Elekta kontaktade Thomas och ville ha hjälp med utvecklingen av en ny strålmaskin kändes det rätt.
Ett annat nytt forskningsområde inom Machine Learning som Thomas kollega Dave Zachariah har börjat arbeta inom handlar om kausalitet - att systematiskt hitta orsak-verkan-samband från mänsklig kunskap och stora datamängder.
- Det känns jättespännande metodmässigt och tillämpningsmässigt, säger Thomas. Att hitta orsakssamband, som exempelvis mellan beteenden och sjukdom.

One of the focus areas of Thomas’s research is medicine.
- It started when I made good contacts with doctors, partly through a leadership course I took. Uppsala is great in that way because we are close to the University Hospital and doctors who conduct research. When my mother became ill with cancer, it naturally affected me personally, so I wanted to find a way to contribute to improving that care. When the company Elekta contacted Thomas and wanted help with the development of a new radiation machine, it felt right.

Another new research area within Machine Learning that Thomas’s colleague Dave Zachariah has started working on is causality - systematically finding cause-and-effect relationships from human knowledge and large datasets.
- It feels very exciting both methodologically and in terms of applications, says Thomas. Finding causal relationships, such as between behaviours and diseases.

Another result from Thomas’s research team is a new programming language. The language is specially designed to give more people access to the possibilities of using the fairly complex algorithms Sequential Monte Carlo (also known as particle filters). With the help of these algorithms, many types of problems can be solved.
- It can be used in virtually all scientific fields, says Thomas. We have recently tested using it in phylogenetics, which involves building trees of how species evolve and become extinct. There, we collaborate with Fredrik Ronquist at the Swedish Museum of Natural History. It has also been used for epidemiological studies, such as the Zika virus.

The common thread in Thomas’s research is dynamic systems, things that change over time and how to mathematically describe them using models that reason with uncertainty. The research can help solve many different types of problems and is highly relevant to most research areas where large amounts of data are handled. There are collaborations with medicine, mathematics, physics, and peace and conflict studies, which have long collected data on conflicts, to name a few.
- The biggest challenges are finding personnel and time. I am constantly looking for competent researchers to recruit

Building a strong team in Machine Learning is Thomas Schön’s greatest achievement, according to himself.
- Five years ago, I was alone, and today we are over 20, says Thomas. From the beginning, I had a strong desire to create a team to work within. I don’t want to sit and research alone; I get bored. Thomas believes that one of his success factors is that he is somewhat of an entrepreneur in the university environment.

- I am good at putting together flexible collaborations, gathering people from all over the world who are best suited to do a certain thing and making sure it actually gets done. The next goal is to try to get more skilled senior-level employees into the team and to disseminate the knowledge so that it can be used by others.

Thomas also teaches, and it is something he enjoys and considers important.

- It is very exciting to create an interest in the students. Additionally, it is refreshing to work with young people who are not constrained but think freely, question, and come up with new ideas. It is a luxury in this job!

”The human role in technical systems is underestimated.”

For computer technology to reach its full potential as part of societal development, we need to understand how people use and interact with technology. How it works in reality. According to Anders Arweström Jansson, Professor of Human-Computer Interaction at the Department of Information Technology in Uppsala, the human role in technical systems is greatly underestimated.

Research in Human-Computer Interaction (HCI) focuses on how to make humans and technology work together effectively. Anders Arweström Jansson wants us to develop computer systems so that those working in traffic control or the process industry have the right computer support when performing their tasks.

- My research focuses on the interaction between humans and computer systems in the workplace. The research includes analysis, design, and evaluation. What do people do, and how can we support them with good design of computer systems? Afterwards, we look at the outcomes and how they can be improved.

Human-Computer Interaction is interdisciplinary research. It does not only focus on technology, as computer science would, nor does it only focus on humans, as psychology would, but it focuses on the interaction BETWEEN humans and technology.

- It is extremely important that the technology, computers, and programs are usable so that various operations can achieve their goals. I focus on usability in safety-critical systems. For example, when controlling trains, things must not go wrong.

- “The human factor” is an overused term that lacks a scientific definition and is somewhat of a Swedish invention. It refers to the human ability to make mistakes, or conversely, the human inability to always do the right thing as an explanation for accidents and incidents. A more thorough analysis shows that the causal relationships are much more complicated than that.

Train traffic control is the main focus right now, but Anders has also researched self-driving cars and cab design in locomotives and bridge design in high-speed ferries. He also works with intensive care and medical technology equipment.

- Mitt intresse för mänskligt beslutfattande i komplexa system kom redan när jag läste psykologi. Jag insåg ganska snart att jag ville jobba med människor i tekniska system och sökte mig till Telia och KTH där jag jobbade med Human Factors-frågor. Jag återvände till Uppsala och forskningen när Institutionen för Informationsteknologi bildades 1999.

Anders’s interdisciplinary research lies at the intersection of Human Factors (a research area focusing on how workplaces, organisations, and the technology used should be designed considering people’s psychological and physical conditions), Human-Computer Interaction, psychology, and cognitive science (an interdisciplinary research area where researchers from fields such as philosophy, psychology, neuroscience, computer science, linguistics, and social anthropology study the nature of human thinking).

Anders Arweström Jansson wants to raise awareness of the importance of cognitive science at Uppsala University.

- Many research areas engage in cognitive science research without making that orientation visible. Uppsala is a blank spot when it comes to cognitive science as a subject, and I want to change that. There should be a cognitive science programme at Uppsala University!

Results from research in Anders’s group include improved methods for cognitive work analysis. One method is Collegial Verbalisation, where both the person being studied and their colleagues watch films of the person working and think aloud about the task being filmed.

- Some knowledge is stored in muscle memory, and such “tacit knowledge” can be difficult to express, but when several people with the same tasks describe the same situation, the data becomes more robust.

The research results are relevant to psychology, computer science, and cognitive science.

One challenge is to align technological development with an understanding of human behaviour. To develop technical systems, one must understand the human role within these systems, which is often underestimated. If a person or organisation cannot use a developed technology, it is wasted money.

Anders mentions the example of self-driving cars, where reality may differ from people’s expectations.

- Either the cars will drive very slowly, or there needs to be someone monitoring the system for it to work. Alternatively, we need to create an infrastructure more similar to existing rail traffic if self-driving cars are to travel at higher speeds.

As a teacher, Anders feels it is incredibly important to spread knowledge to students who are the next generation of clients and developers, and those who will create the future.

- Human-Computer Interaction is a sought-after subject in many programmes today, so the outlook is promising. Looking back twenty years, a lot of positive changes have occurred.

Interview by Kajsa Örjavik, 2018-11-07

“We must digitalise with care.”

How do we best introduce digital technology into society and our lives? According to Åsa Cajander, digitalisation must be allowed to take time and cost more in the development and implementation phases to avoid creating new problems.

- I want to ensure that people’s work remains beneficial for them even as it becomes digitalised, and I want to help with methods to achieve this.

Computerised information systems, web and app-based e-services, teaching via digital learning platforms – the influence of information technology on our lives is only increasing. New electronic tools promise to make our lives more efficient and easier. But how well do the ambitions match the results? Researcher Åsa Cajander studies how the digitalisation of various societal functions affects work and individuals.

- Today, we are creating digital divides in society and people who are burnt out. Digitalisation follows the same curve as the burnout curve. It is not the only reason we burn out, but it is one of them.

The underlying factors may seem overwhelming, as numerous and complex as the individual users. But Åsa Cajander believes that IT system designers must still better understand the interplay between technological development and society at large.

- As it stands, those working in systems with technical projects find workplace, gender, or cultural issues so incredibly messy that they don’t even know where to start. But we can’t think like that. Just because it’s complicated doesn’t mean we can ignore the problems. We must still do something to create digital systems that work better for us than they do today.

As a researcher in Human-Computer Interaction, Åsa Cajander uses so-called action research. This research involves collaborating with various societal groups in projects to contribute to improvements and participate in change efforts. The focus is on interviews and surveys with participants, as well as meetings and workshops.

- Then my research group and I often write articles together with those involved in the project. So, you are not a fly on the wall like many researchers are.

One of the largest projects Åsa Cajander has been involved in was when Region Uppsala introduced electronic patient records in 2011. She was responsible for studies with patients and healthcare staff before and after the controversial implementation.

- The healthcare staff felt that the change was very top-down. Even today, many of them find it very difficult to see the positives of online records. They believe it creates anxiety among patients and also greater time pressure for staff when patients contact them with questions about what they read, says Åsa Cajander.

On the other hand, 97 percent of all patients think it is obvious that they should be able to access their records online.

- We have a problem with continuity in healthcare. Now at least the patient can log in and follow their care, see where the referral is, when the next visit is, compare test results, and so on.

One of the challenges is whether e-records should be made available in languages other than Swedish. Complex words and technical terms should have links to glossaries. Additionally, doctors rarely mention the possibility for patients to read their records online and thus prepare for their visits. This could give patients a better understanding of their illness and greater motivation to follow the doctor’s recommendations, says Åsa Cajander.

As a researcher, she can make recommendations but rarely influence decision-makers directly - although it has happened.

- In the e-record project, there was a discussion about a two-week waiting period before patients could log in to see test results and read what the doctor had written. My research group conducted an interview study with 30 cancer patients, asking how they wanted it to work. And they did not want a two-week waiting period! The study contributed to the decision not to adopt the proposal. There was resistance from the doctors, but they had no choice.

She also points out that it is necessary to create good conditions for the administrative work and work environment in healthcare. Doctors’ health is deteriorating, and the number of nurses is decreasing, while the aging population needs care for an increasingly longer part of their lives.

- The equation doesn’t add up. And the biggest problem is not the money but the healthcare staff and their numbers. We need to work to ensure they enjoy their profession and stay. Therefore, we must also digitalise with care.

She adds:

- It is very much about change management and communication. Communication is fundamental; without it, everything falls apart.

Her research group consists of twelve people from various fields – one PhD student is a cultural geographer, another is a computer scientist in Human-Computer Interaction, and a third comes from the engineering programme Systems in Technology and Society. The senior researchers in the group have backgrounds in gender studies, business administration, and computer science. With a research leader specialising in work environment issues, it is no wonder that interdisciplinary collaboration works well.

- I work a lot to create good conditions for collaboration, and we have, I would boldly say, an incredibly good work environment. We get along very well together. There are different opinions, but also a great deal of respect for each other’s expertise, says Åsa Cajander.

She herself is originally a language teacher in French and English. In her work at Fryshuset in Stockholm, she met many young people with psychosocial problems, “there was a lot of psychology, behavioural science, and conflict management.” But then her interest in mathematics and computer science took over, subjects she had studied alongside her language teacher training at Uppsala University.

- So in 2000, I switched to being an IT consultant at a large international IT company. I worked with Java programming in the industry and taught courses. At the same time, I took evening courses at the university because it was so much fun!

These courses included programming, systems development methods, and teaching and learning, before I finally took the step to become a PhD student in Human-Computer Interaction. The winding path to higher technical studies is an experience she shares with many women, says Åsa Cajander.

- For women, it has been observed that when they end up in technical professions, this winding, unusual background is quite common. Therefore, we must support and find ways for students to switch to technical subjects later in their academic careers. Today, there are not enough bridges between educational programmes.

Åsa Cajander currently has a three-year mandate as an equal opportunities representative at the IT department. Additionally, she conducts research on e-health and digitalisation within the Nordwit centre, which is run from the Centre for Gender Research. Besides a handful of other research projects, she teaches 30 percent of the courses in Medical Informatics, IT in Society, and Complex IT Systems in Large Organisations. And she blogs. And tweets.

How do you manage, and above all, have the energy for everything?

- I think it’s because I research and lecture a lot about the balance between work and leisure, and the symptoms of stress. I know what I can handle and have the tools to slow down in time when it gets too much.

- Then I’m good at resting, haha. And I sleep a lot, 9 hours a night.

Interview by Anneli Björkman, 2018-06-05

His algorithms replace uncertain assumptions

Is it possible to ensure that patients receive the correct dosage of medication? Or take their medicine as prescribed? Alexander Medvedev’s mathematical data models replace uncertain assumptions with automatically made decisions.

In healthcare, the need for modern and effective treatment methods and techniques is increasing. Computerised control systems play an important role in planning, decision-making, implementation, and follow-up of treatment and medication. The foundation of automatic control is control theory, a research area where Alexander Medvedev is one of the model designers. His focus is on how regulations work in living nature, including the human body.

- Generally, I try to build mathematical equations to explain measured hormone levels in the blood, says Alexander Medvedev, Professor of Automatic Control at the Department of Information Technology. I also use mathematical modelling to optimise and personalise treatments, including automatic dosing of medication.

The work is carried out in collaboration with researchers in neuroscience, medicine, and pharmacology, as well as doctors at Uppsala University Hospital. The collaboration with Uppsala University Hospital began in 2007 when a doctor approached Alexander Medvedev with a problem regarding the regulation of endocrine systems.

- At that time, it was about women with premenstrual symptoms, and the doctor was a gynaecologist. I started digging deeper into the problem and discovered that there was no coherent mathematical theory for how the brain regulates endocrine glands, which is what needs to be addressed.

Alexander Medvedev then started a project on how to build mathematical models for endocrine systems. The research team’s theory to describe how the brain controls endocrine systems with hormone impulses is considered his most significant research achievement to date.

- The success lies in the fact that we started from scratch and arrived at something that matches experimental data and biological facts. I am currently writing a book about this together with my Russian colleagues. My collaborations with doctors at Uppsala University Hospital have continued, including work on automatic dosing of medication during anaesthesia or for diseases such as Parkinson’s.

Today, there are several ways to address issues with self-dosing using apps on mobile phones, tablets, or computers. These apps remind patients when their medication needs to be taken or how to adjust for missed doses. The software can also help measure symptoms and compare them with the patient's history, says Alexander Medvedev.

- Medication can be very difficult to assess on one's own. According to doctors, patients tend to overdose, which is not good either, as increased dosing increases complications, for example, in Parkinson's disease."

Constructing equations that describe medical treatments, he admits, is a challenge both mathematically and engineering-wise. An important part is being able to measure the patient's symptoms objectively, and for that, computerised technical solutions are needed. One such technique is called eye-tracking, which is used in Parkinson's care. It involves the patient observing and following a stimulus in the form of a dot on a tablet while two built-in cameras read the eye movements. By comparing the eyes' reactions to the dot's movements, doctors can assess how severe the patient's neurological symptoms are. According to Alexander Medvedev, such technical aids provide a more reliable basis and a higher level of efficiency, saving both time and money.

- To determine which values can be neglected and which are important, I start with mathematical analysis. Then I go to the biological or medical data available and compare what I see in the equations on the computer with the data published in biology and the signals that doctors can measure. Then I start thinking about whether we can do an experiment. That's when it gets exciting because we test a hypothesis, and for that, collaborations with doctors or biologists are needed. That's the hardest part."

His latest focus is deep brain stimulation, a technique that has been used since the 1980s for Parkinson's disease but also for epilepsy, essential tremor, and other neurological conditions. The treatment is carried out at Akademiska Hospital and involves implanting electrodes in the brain, which are then stimulated with electrical signals. As a patient, you get a remote control to turn the stimulation on and off and switch between different settings. The other, more advanced unit is used by the doctor to program the stimulation. What Alexander Medvedev and his research team have done is develop mathematical models that calculate the most optimal settings for electrical stimulation.

- We look at MRI scans that show what the patient's brain looks like and where the electrodes have been implanted. Then we can calculate how the electric field spreads in the brain and thus how the stimuli should be designed to cover a certain area without spilling over. This way, adjacent areas are not affected by the electric field," says Alexander Medvedev.

- There can be several settings for the stimulation that give similar results according to our mathematical models. The doctor can then try these on the patient and measure the treatment's effect on symptoms such as tremor and eye motor skills. Such a study is currently underway at Akademiska Hospital. As far as we know, this is the first time this is being done in Sweden.

The team's first study has shown that their mathematical models correspond quite well with reality and that the stimuli calculated by the models work on patients. On average, they have achieved a reduction of more than 40% in medication due to brain stimulation. This is particularly important in Parkinson's disease, as the difficulties in dosing correctly increase with the progression of the disease, explains Alexander Medvedev.

- Some patients become almost symptom-free in terms of motor skills with the stimulation. However, there can be side effects if the wrong area is stimulated, as different parts of the brain are responsible for different functions. Therefore, it is crucial to calculate where the stimulation is located, and doctors cannot do that today. They cannot see into the brain, but the computer can.

More about Alexander Medvedev och his research

Interview by Anneli Björkman 2017-08-30

She finds numerous solutions to physical problems

She loves to delve into mathematical algorithms but needs to know that her models will be of practical use. Through calculations of physical phenomena, Gunilla Kreiss aims to bridge the gap between ideas and reality.

- When I do numerical calculations on the computer, numbers do come out, but they don't become numbers that have anything to do with reality until someone else uses them."

The professor of numerical analysis becomes enthusiastic when she describes how she stumbled upon her research subject. During her engineering studies at KTH, she was so captivated by numerical analysis that she never wanted the courses to end.

- I felt there was a connection between 'here is a real problem, now we make a model and then we solve this mathematical problem that is the model.' I liked that you could actually see that the chain was there, says Gunilla Kreiss.

- What I also found exciting was that experiments were moved into the form of these numerical simulations on the computer instead of doing measurements in a lab.

- Today, it is possible to quickly and efficiently solve various types of computational problems with the help of computers. Mathematical problems can be transformed into numerical computational models in many different ways. These are tested and combined from countless perspectives to find the best possible computational model. Despite all the efficiency, the challenge of analysing the methods used remains, says Gunilla Kreiss. The computer has not completely replaced the blackboard.

- Writing things on the board, that's great stuff!" she laughs. "And bouncing ideas around, talking about things, those are the ideas you talk about. Then the actual calculations can partly be done with analysis. I often use classical mathematical analysis."

In addition to computer simulations, the methods developed in numerical analysis or computational science can be applied to problems from many different areas. Gunilla Kreiss researches physical phenomena that can be modelled with partial differential equations, and in such a project, she collaborates with physicists at Uppsala University.

- My role is to develop and analyse simulation tools to be used for creating small magnetic information carriers like computer memories. Magnetism is crucial for storing and transmitting information. Ultimately, it would be very exciting if I could contribute to even more efficient computer memories, says Gunilla Kreiss.

The project is one of several she leads within the framework of the strategic research initiative Essence of e-science. Another project involves developing computational models with geoscientists using simulations of elastic waves in the Earth's crust. This way, geologists can predict scenarios during earthquakes in different geographical areas.

- In both collaborations, there is a reality that is not so far away, and where I see that what I work on can make a difference. But the benefit doesn't have to be immediate; it can be a few steps away. That's perfectly fine, as long as I can see that there is a connection.

But being able to see the logical consequences is not the hardest part, according to Gunilla Kreiss. It's formulating the questions.

How do you go about formulating the best questions?

- It's difficult! It's about turning things around and trying, and coming to 'no, I can't answer this,' or 'no, this is too easy.' Because that's how it is. Maybe you start with the hardest question, and then it doesn't work, you don't come to a solution. Then you have to simplify. At the same time, you have to try to find a balance so that you don't simplify away all the essentials.

Interview by Anneli Björkman, 2016

She has a feel for sustainable things

Computers, mobile phones, and remote sensing devices are forming ever-larger networks, enabling increasingly detailed, innovative, and efficient interactions. In her research on wireless sensor networks and mobile applications, Edith Ngai focuses on optimising communication and the quality of information from raw data.

- The most important thing is that this data can be processed and collected in a cloud with greater capacity for storage and computation. Then we can gain knowledge and insights that we have never had before, and perhaps improve the world in ways we never could before.

In 2014, her research group received funding from Vinnova for the Green Internet of Things project in collaboration with Uppsala Municipality, KTH, Ericsson, IBM, and four other business partners. The aim is to provide Uppsala with a wireless infrastructure of sensors that collect environmental data for sustainable urban and transport planning. The network will help monitor energy consumption and reduce traffic congestion, says Edith Ngai.

- The idea came the year before, when we conducted a project with some Chinese universities on pollution problems in Beijing, where the situation is naturally much worse. In Uppsala, the poor air quality is largely due to the studded tyres we use in winter, which scrape against the asphalt and cause particles to swirl up. Since these particles can cause health problems, we are very keen to measure them with our sensor technology.

Paving the way for the smart city of the future requires interdisciplinary collaboration and includes specialists in areas such as air pollution, traffic, urban planning, and decision-making. Edith Ngai also pushes the boundaries of the scientific field in another e-health project where she collaborates with researchers in psychology, psychiatry, and biology.

- I am interested in how people's quality of life can be improved and needs met, whether it concerns children, young people, or the elderly. That is how information technology should be used.

The mobile application SADHealth is a tool to monitor how light exposure affects activity levels, health, and mood. The main goal is to provide users with a program that can identify and help them understand the symptoms of Seasonal Affective Disorder (SAD), says Edith Ngai. She came up with the idea during her first winter in Sweden.

- Since my hometown Hong Kong is closer to the equator, the change between summer and winter is not so noticeable. But when I moved to Sweden in 2008, I noticed how long the days were in summer and the complete opposite in winter. During the darker season, I needed more sleep and had less appetite. And when I went to Hong Kong during Christmas, my skin looked much paler compared to other Asians. This made me reflect on how seasonal changes can affect people.

SADHealth cannot yet be used for medical purposes but only as a tool to measure how much light and exercise the user gets. Today's smartphones are equipped with all the sensors needed for the SADHealth app: camera, thermometer, light and proximity sensors, GPS. Several of Edith Ngai's master's students are developing the application as part of their theses.

"We are continuously improving the app and plan to make it gamified so that friends can compete in how much exercise or sunlight they get in the summer. This way, people can encourage each other to be more active and spend more time in the sun."

She also teaches courses in computer networks, "basically everything about how computers talk to each other." In the computer science project course, students work for a term with a company, currently Ericsson Research. Since February, she has been gaining practical experience as a guest professor at Ericsson.

- My task is to secure data, which many believe is becoming increasingly important. If you want to transfer data efficiently, you need to have many routers cooperate in the network. But the question is how to achieve security while taking advantage of efficient and broad data distribution?"

So where do such security measures need to be implemented?

- Probably in various places. We are investigating this problem and looking at whether the end user can control how data is decrypted or made available. Collecting a large amount of data involves a potential risk of revealing personal information. If we cannot protect personal privacy when collecting data, people will be too afraid to use, for example, health applications. This balance must be resolved.

Interview by Anneli Björkman, 2015

She Assembles Life-Saving Pixels

Today, it is common for biomedical experiments to result in thousands of microscope images. To extract and interpret information about the conditions in cells, tissues, and organisms, automated image analysis methods are required. In a few hours, computers deliver analyses that would take a human several years to produce manually, according to Carolina Wählby.

- With digital image analysis, my colleagues and I can quickly measure how cells react to a large number of treatments. You can't test 250 different drugs on a patient, but we can take the patient's cells and culture them in small wells in plastic plates. Then we add different drug substances to different wells and image the cells through a fluorescence microscope, she explains.

- We already see that known drugs used today have different effects at the cellular level, and sometimes different effects on cells from different patients. To measure these effects, we need advanced software.

The computer programs Carolina develops are based on mathematical algorithms that can identify objects and measure shape, colour, and patterns. Here, observable properties are converted into numbers that reveal connections the human eye can easily miss.

- We try to quantify changes in cells and tissue in various disease states, from cancer to skin changes, and how UV light affects the cells in the eye's lens. In one of our projects, we have also developed new image filtering algorithms and methods to quickly and efficiently find image objects of a given size.

But no matter how effective and reproducible the methods are, they still involve compromises, says Carolina Wählby.

- It is difficult to get both data and analysis perfect, largely due to biological variation. The methods must not be too sensitive. Even if you have cells that are clones of each other, they are not identical. Nature itself is not ones and zeros.

It was a desire to measure cause and effect that once led her to the field of image analysis. After studying the engineering program in molecular biotechnology in Uppsala, she did her thesis at Karolinska Institutet in Stockholm. There, she cultured cells that were photographed under a microscope.

- One day my supervisor said, 'now we need to get some numbers too, you can print the images on paper and then ask the lab assistant to count the dots from the images.' But I didn't have the heart to do that - the lab assistant was a wonderful older lady whom I had immense respect for. Asking her to sit and count dots was not in my world, laughs Carolina Wählby.

She decided to learn digital image processing herself and enrolled in a course at the Centre for Image Analysis at Uppsala University. This was followed by a doctoral position in digital image analysis with Professor Ewert Bengtsson at the Department of Information Technology. Today, she leads her own research group at the same department.

- Our laboratory works extensively with image data produced at SciLifeLab. Additionally, we have a dozen other collaborations with researchers around Sweden and the rest of the world. We work with model organisms to test drugs on a large scale. One of our organisms is the 1-millimetre-long roundworm Caenorhabditis elegans, which we use to find new anti-infective therapies and genes involved in fat metabolism.

Many of Carolina Wählby's research contacts were made during her six years at the Broad Institute of Harvard and MIT in Cambridge, Massachusetts, USA. In her role as project leader, she had the opportunity to help develop what she considers her greatest contribution to the research community, CellProfiler.

- It is software that can also be used by biomedical researchers without an IT background to measure and analyse microscopic changes at the cellular level. Just finding the cells, the first step before you can measure them, is based on my doctoral thesis.

CellProfiler only requires a PC or Mac and is free to download and use. Since the source code is completely open, anyone is free to build on it.

- Making these methods so accessible to so many is very exciting. Every day, more than two new scientific articles are published using the tool. In this way, we broadly advance research and support many different scientific fields.

In another notable project, her research group in Uppsala collaborated with Mats Nilsson's group at SciLifeLab and Stockholm University. They have demonstrated a new way to measure where different genes are active in a tissue. To visualise gene expression at different image resolutions, the researchers use tools similar to Google Maps, where gene expression can be shown or hidden in the same way that roads and road names can be shown or hidden when looking at aerial photos, explains Carolina Wählby.

- It becomes particularly interesting when you want to find out the differences between normal tissue and cancer and learn more about which types of cells, such as immune cells, are active at the surface of a tumour. The tool, which goes by the working name Tissue Maps, makes it possible to study the results at different resolutions and choose to display different types of information as layers on top of the original image.

She concludes:

- It can provide important answers regarding individual variations in, for example, cancer tumours, and hopefully lead to a better understanding of the disease process and ensure that patients receive the best possible treatment.

Interview with Anneli Björkman 2015