Bank hacks Healthcare

Blog -

Why does cancer treatment prove effective in some cases but not in others? Can we use data to predict whether or not a treatment will work? In collaboration with the Erasmus MC and Princess Maxima Pediatric Hospital, Tjebbe Tauber introduced this moving and exciting experiment as a topic for the Beyond Banking Days hackathon. The outcome was surprising. In his blog, Tjebbe explains how bankers tackled a healthcare problem.

I was introduced to thoughts like "ABN AMRO has to embrace open innovation". Tjebbe Tauber IT Service Level Manager

From an IT perspective, banks and healthcare institutions have much in common. Security is a top priority, as client or patient data must be safeguarded by applying the highest possible privacy standards. And both types of institutions have a lot of legacy IT, a mass of dated, interconnected and mutually interdependent IT systems. If you change something in one system you might accidentally disrupt functionality elsewhere. Banks and healthcare institutions have some major differences too, however. Where banks generally have a long history of big investments in IT, innovation and data analysis, the healthcare sector lags behind in this area.

Beyond Banking Days

After the Erasmus meeting, my next goal was to find colleagues who wanted to help organise a hackathon, so we could flesh out the new idea. Thanks to @Jip Tauber (family ties coming in handy) and @Jarno Zange, I ended up talking to @Micha Rentier and the colleagues with whom I would later join the Beyond Banking Days (BBD) team. During ABN AMRO's Beyond Banking Days, the team aimed to organize the bank’s biggest hackathon ever, our aspiration being to redefine the financial sector's future. It's a smart, ambitious group of people just bubbling with fantastic ideas for what the future might bring. I was introduced to thoughts like “ABN AMRO has to embrace open innovation”. Clients and external developers have to be involved in the development of our services. It's a little bit scary, of course, to give them a look behind the scenes. After all, there is no way to predict what kind of prototypes these people will create and what they’ll make of the banking technology that's made available to them. But the BBD team believes this method paves the way to great new products and ensures that the internal and external community developers will gladly build on our services rather than our competitors'. Working with these colleagues has been a very inspiring experience.

Social impact challenge

The Beyond Banking Days hackathon of June 2017 was to include a social impact challenge. Following the Erasmus discussions I was given the opportunity to design this challenge, for which I collaborated with Peter van der Spek (Erasmus Hospital) and Jules Meijerink (Princess Maxima Pediatric Hospital). Through them, I discovered that the healthcare sector has seen an increasingly strong focus on big data, ever since the genome project was completed. In the genome project, which was about mapping the human genome, we learned that human DNA consists of 3 billion “characters”. If any of those characters get mixed up or damaged, the damage could manifest as cancer. Those insights are still relatively new and the healthcare sector remains in the early stages of further analysing and applying them. Besides, data analysis has not yet become part and parcel of the healthcare sector as fully as it has in the banking world.

The challenge

So, the question was what we might learn if a group of clever data analysts from the financial sector examines a genetic dataset of children with cancer. To word it more concretely – “Let's take an anonymised genetic dataset of 270 children with T-ALL leukaemia, and then look at the treatment results in light of some questions. Why does treatment prove effective in some cases but not all of them? And can this DNA tell us which gene pool causes that variable? If so, can we use DNA to predict whether or not a patient will respond to the treatment?” The scientists hypothesized that it should be possible to make predictions this way. If they could successfully determine the genes involved in treatment response, treatment plans could then dramatically be improved based on those insights. After discussing it with people at Erasmus and Princess Maxima Hospitals, we introduced this moving and exciting experiment as our topic for the social impact challenge at the BBD hackathon.

The successful hack

A total of 41 teams participated in the Beyond Banking Days hackathon. Four of those teams ran analyses on the DNA dataset, and the results far exceeded our expectations. The teams managed to identify the gene pool that determines whether the relevant treatment for children's leukaemia will work! This positive outcome sparked in the four teams a sense of euphoria and pride, which then spread to the other teams and participants. Of course Peter and Jules, who were both present, shared in the enthusiasm and excitement.

Moving forward

In the meantime, a PhD student has been engaged by Prinses Maxima Hospital to validate the model the teams created. If that's a success, Jules Meijerink's research group will investigate ways to help patients for whom the current treatment doesn't work well. That's not all, though, as more good things have come out of this initiative. Last September, Erasmus Hospital added some of the analysis methods used by the teams to its own doctors' curriculum. The Erasmus is also working on another medical problem, for which they’re now creating a new dataset for analysts to examine. 

All in all, the hackathon delivered fantastic results with unexpected and exciting spin-offs. And we’re already looking ahead. Next we have the 2018 edition of the Beyond Banking Days to organise and look forward to. We intend to come up with another inspiring, impactful, innovative healthcare project to enter into the hackathon. I just can't wait!

Share

Join the discussion

ABN AMRO would like to know your opinion, so below this article you can react to this article via Disqus. By doing so, you agree to the conditions for reacting to articles on our website.

More blogs