16 Aug 2019Blog

Why a unique solution requires a unique way of working

Mari Karsikas Solita

There’s a lot of hype around the transformative power of data. But what’s it like to actually work on a solution that harnesses data to bring business benefits and change people’s lives for the better?

Developed in partnership with Coxa, a leading orthopaedic hospital, Oravizio utilises data from over 35,000 surgeries to accurately predict and assess the risks for patients considering joint replacement surgery. It also helps doctors to communicate effectively with patients about their treatment. A further benefit? It reduces the risks of revision surgery, saving costs for society.

As a concrete clinical tool that has gained CE certification, the development work for Oravizio was a unique project.

Mari Karsikas, Data Scientist at Solita, was at the forefront of this development. We spoke to her about the challenges and rewards of working on such a complex project.

Hello, Mari! How did you get involved in the Oravizio team?

Hello and thanks for asking! Already before I started Coxa had already carried out some preliminary studies with the data they’d compiled. The results were so promising they decided they needed to utilise the data better. At this point, I joined the project with my Solita colleagues to develop machine learning solutions.

And what was your role?

I’m working as a data scientist in the Oravizio team. The data scientist is the person in charge of the risk models in the background of the tool. The role includes things like preprocessing, statistics and visualizations, data subset optimization, and training and testing the machine learning models.

It’s been an engrossing and fascinating experience to create the prediction models for this very relevant need with a unique data set. And the role isn’t restricted only to the models. It’s important to understand the whole chain from data storing to the frontend of the tool, because a small change or update in the data or in the tool often affects all the layers.

Oravizio is a unique solution. Did development pose any particular challenges?

The project’s been a stream of challenges but in a positive way. So much has been new to us. We haven’t developed a CE marked medical device before. For me, the most challenging thing has been regularisation issues. And from a data science perspective, one big challenge has been how to get enough good external evaluation data sets. But now we have many good contacts and things are going forward. None of the challenges have been insurmountable.

Developing a CE marked medical device also requires a clinical evaluation process, which means a lot of documentation work and writing for scientific publications. It’s interesting but also a little bit funny, because when I left my work at the University of Oulu, I thought I’m free of scientific writing. And then I found myself in this project!

And how about the implementation process?

From my perspective as a data scientist, it’s been a little challenging that we started data research and software development almost at the same time. The idea was that if promising models were to emerge from the data, we could also have other parts of the tool ready sooner, saving time overall.
It means we built  the tool for the user even before we had final results from the data or final machine learning models. But in retrospect, it’s easy to say this was a very good idea. We took a risk, but we saved time.

What was it like to work with a multi-site team?

Working multi-site is the usual way to work at Solita and we have many ways to communicate remotely. In this project we have people from Tampere, Helsinki and Oulu. Anyway, we are all work mates in the same time-zone so what’s the problem? One of our team members is dreaming about a Solita office in New Zealand and I believe it would work too.

Some rare times, when wi-fi freezes, video disappears and my voice sounds like a robot to my workmates in a daily meeting, I can feel like a lone wolf howling at the moon. But then I jump onto a train and travel to meet my colleagues as I do every second week anyway. And work goes on again.

And how did Oravizio differ from other projects you’ve worked on?

I’ve been at Solita for two years. Before that, I worked in the university. With Oravizio, we’ve developed a tool which is a product and a medical device. That’s a step further than in the academic world, where the main goal often is to produce new information for the world. Now we do that but also utilize the new information for a concrete product in daily clinical use, and that’s exciting!

Oravizio also differs as a project from my other Solita assignments because it’s not only traditional customer work but also Solita’s own product developing project. Anyway, regardless of the business domain, my work usually has the same features: we store, transfer, process and analyse data.

Finally – what do you enjoy most about your life at Solita?

The simple reason I’m working here is my workmates. They are very professional, inspiring and humorous. They are great people, all of them.

Would you like to work on complex and challenging projects like Oravizio that combine data science, software engineering, and design competencies? We are always seeking diverse talent, so check out our open positions.