It is the discoveries, it is bringing people together, I get to learn something new every day. I get to learn a little bit about healthcare, a little bit about languages or linguistics, a little bit about statistics and a lot about computing. I love when we translate research from universities to real life practice and see the changes.

Can you introduce yourself for us?

My name is Hanna Suominen. I'm an associate professor in the ANU School of Computing.

What do you study?

My topic is machine learning for healthcare, and in particular natural language processing applications.

I'm actually quite passionate about real life impact, making our society better, improving healthcare, releasing healthcare professionals’ time from writing notes or documenting their work gathering data, to spending time with patients, doing clinical caring, doing those things that people do the best. I can contribute my research time, my talent in machine learning and mathematics towards creating applications in this field.

How can machine learning and natural language processing help in healthcare?

So one example is natural language processing to record conversations in hospitals. We can attach microphones on nurses while they have shift change handovers. When a team of nurses is ready to leave work and other nurses are coming in, they normally have a discussion by the patient's bedside about what happened during the shifts. We can capture that conversation, convert it to text and fill out forms with that information. So that releases nurses’ time from capturing that conversation. It also can contribute to reducing errors because basically, the information is coming there straight away. Another thing is diagnostics. So nowadays, we gather so much information about patients and there are thousands and thousands of new papers being published every day. It's really difficult for clinicians to keep up with all of this information. And the same is true for patients themselves. So search engines finding the related evidence, and not only from published literature, but also from the patient record, summarizing it, synthesizing it, suggesting what could be possible ways to go forward and why. Overall, these are decision aids, decision support systems for situational awareness.

Where did you grow up and how did you come to be a researcher at ANU?

I moved to Australia over 10 years ago. I was born in Finland in a little countryside town called Loimaa in the Southwest corner. I did my schooling there. And after year 12, I moved to Turku, which is a regional capital there in Southwest of Finland. I studied in the University of Turku, I studied mathematics, with my specialty in mathematical modelling related to medical imaging, so positron emission tomography. And after that, my Masters which was a research Masters. After that I did a PhD in the same city, same University of Turku, about clinical text mining. And after that I moved here, I did my postdoc in the National Information and Communications Technologies Australia, NICTA, I studied wearable sensors mainly around supporting our elite athletes in performing better. And after that I joined ANU four and a half years ago now and my duty now is to lead the OHIOH Grand Challenge which is Our Health In Our Hands, to bridge health scientists, clinicians, patients themselves or family members of the patients and us, computing and engineering experts. We are putting our health in our hands literally. So apps, minimally invasive techniques, things that can be used at home so you don't necessarily need to live in a city where you have a big hospital. You don't necessarily need to go to a medical imaging department. We could possibly do something based on these monitoring applications, which can be thought of a bit like combining my postdoc on swimming sensors assessing what people are doing when they train for swimming and combining that with the clinical text mining.

What do you find most exciting about the work you're currently doing?

It is the discoveries, it is bringing people together, I get to learn something new every day. I get to learn a little bit about healthcare, a little bit about languages or linguistics, a little bit about statistics and a lot about computing. Most recently, I have been studying more and more about cybersecurity. So how to make these systems safe to use, rather than just making the decision support accurate or correct. Not only in machine learning, but also fields within software engineering. I also love when we translate research from universities to real life practice and see the changes there.

How does supercomputing and NCI play a role in your research?

So in machine learning, or computational science, or data science, we tend to have what we call big data, so big data sets, they need to be stored, they need to be processed. And most importantly, they need to be evaluated as high quality systems. So for example, this decision support needs to be accurate, we cannot miss something. And we cannot either generate extra alarms. It’s not only the storage and processing of the data but the evaluation step that’s important, and then going back and improving the processing pipeline so that we can get better and better quality. That's very computationally heavy. This processing, evaluation and even sometimes the data it cannot fit on a normal tablet or laptop, or office computer. And even if it could, we would be running experiments for years and years. So what the supercomputer provides is this processing capability. And we can run these experiments in parallel, very, very fast, and then make discoveries faster.

What has Gadi helped you achieve?

I'm having the privilege to lead a team of approximately five researchers and 15 PhD students. And in addition to that, we have Bachelors, Honours and Masters students. So it's a quite sizable team at the moment. And what we can do with Gadi is collaborate. So not everybody needs to do the same task over and over, we can set up these processing pipelines and apply them from a project to another. That's one advantage. And of course, the other advantage where diversity comes into play is that sometimes in these experiments, what we want to do with a diverse team, is to use everybody's time and kind of have discovery, like will this work? Will that work? Will this improve the performance? So you can have a go and have much faster discoveries when people do at the same time independent tests, and then we bring it all together. So it makes things a lot faster. And finally, it's a stable environment, in a sense that we're in OHIOH, we have this partnership across not only disciplines in a venue, but for example, ACT Health, we can prepare our way for the future and have a stable environment then take steps towards having everything ready for the final translational research output when there's a stable environment. So I would say that it's mainly about being able to do things a lot faster. But it's also a way of making people work together and making things a lot easier. So it's not only about computing, it's also about people.

What do you want people to know about the work you do?

At the moment, we are in this Grand Challenge, a new strategic initiative worth $10 million. We are now about two and a half years in of this five-year initiative, we have moved on from preparing the project and having our initial searches, we now know where the niche is. And we are publishing original research and making real improvements in health research, and in computing, research engineering. It's really exciting. And it's also exciting that we are getting more and more participants joining our cohort. We work on diabetes and on neurological conditions, mainly MS. We are recruiting participants, and it's so exciting that the data is coming together. We can work on that data and test our machine learning pipelines, engineering pipelines, on those new novel data sets.

What are you most proud of in your career?

Around 10 years ago, when I was talking about medical machine learning, machine learning for healthcare, natural language processing, speech recognition in this context, and these wearable sensors, people said, “Hanna, you have gone nuts. That's way too hard, it will never happen.” And look at where we are now: we all are using our sports sensors, we are all applying telehealth, we have electronic health records. We have decision support systems happening in hospitals. We have been able to release clinicians’ time, we have been able to inform patients better and empower them. It's all happening. And I'm so excited that we have been bridging the gap. And I didn't go nuts. It's here and it's happening.

How does NCI benefit the Australian research community?

Computing is something that enables discoveries in every area of society, from banking to self-driving cars to now healthcare. Without a computational infrastructure like NCI, we cannot make discoveries, we cannot have the potential of these data sources and the raw data that is out there, convert that into information, knowledge and wisdom. It's an infrastructure that we need, then also it brings people together. Not only in Canberra, where I'm based, but also with, for example, Sydney and Melbourne, and even internationally. For example, in the pandemic times one of the OHIOH PhD students is studying from Italy. So even that is possible and it's a lot simpler to manage when it's a centralised facility.