Medical, Computer Science, Design, Biomedical engineering, Business. I collaborate with so many different people with different backgrounds on a daily basis that there’s always someone to ask, “how would you approach this?”
Thomas Hagebols is a Machine Learning applications intern, working on making AI available on small handheld devices, out of the Philips office at the High Tech Campus in Eindhoven.
In this interview, he talks us through his experiences with machine learning and why it holds such a fascination for him, his journey and experiences so far at Philips and why he’s passionate about using his IT skills to contribute and help medical professionals.
As a final project for my degree, I opted for an internship at Philips. Machine learning and AI is booming. I wanted to make something that people would really use in their everyday life or that would help doctors in a hospital achieve more. It’s cool to work on stuff that matters.
I studied Industrial Engineering at Eindhoven Technical University. After my bachelor I started a masters degree in Business Information systems (BIS). During my master AI started to emerge and really cool stuff seemed to be happening. Machine learning and AI is fascinating as it is a cross over between Computer Science, Neurology and Psychology. I didn’t want to miss that chance and I decided to pursue a second master Data Science with a strong focus on deep learning. To gain experience I started looking for an internship. Via referral, I knew someone at the High Tech Campus and jumped at the opportunity to work at Philips.
I’m a Machine Learning intern at Philips. I work on making Artificial Intelligence available on small devices. Efficient AI can make devices smaller, cheaper and more user friendly and is really exciting for healthcare as well as everyday life.
You have to be very careful with your data as it’s all about what you train your AI on. A common mistake is that the data is not diverse enough. Say you want to detect a disease that is common among older people, but can occur with young people as well. If your training data only contains data of older people the AI might fail to recognize the disease with a young patient, because it never saw a sick young person before. I think Philips is one of the few companies that knows how to handle these subtleties and that’s what makes them such a strong player in the medical field.
AI has the potential to really support diagnosis. Think of it as an automated second opinion. It’s a positive tool in the process, not a replacement of process.
One of the biggest opportunities for AI, is to really help doctors with diagnosis. If you’re taking a biopsy, for instance, or in an MRI, you’ll be looking at image on a microscope for a long time and at some point, fatigue will set in. AI can make a positive impact to these types of repetitive tasks, highlighting elements to support the diagnosis.
An application of AI might be a handheld ultrasound device that a doctor can use on the go. This ultrasound could automatically detect suspicious tissue to help the doctor with diagnosis. Another application might be a bracelet that detects when an elderly person falls and automatically raises a call for help.
It’s exciting to be working at the frontier of technology, working for a company where I can do something beneficial, not just focus only on making lots of money. For me, it is about discovering that in healthcare, I can make a real impact. I can help indirectly by making something really useful.
At the moment, I’m working on a new kind algorithm and that is inspired by how the brain works. We figure that the best example of intelligence is our human brain, so why not try to recreate it? We are looking at the structure of the brain and try to recreate how neurons connect, grow bonds and disconnect. We think that more closely resembling the brain helps improve our current algorithms and makes them better and more efficient.
If you don’t pay attention, watch out! Most tools I use today didn’t even exist 3 years ago. Things change FAST. I remember some guys inventing a bot that could speak text that was almost indistinguishable from real human speech. It took a few hours to generate just a few seconds of text. A year later this was sped up 1000X and it now takes seconds instead of hours.
I’m particularly fascinated by the impact of AI on human technology interaction. How can use AI in a way that people trust it and want to engage with the technology. I’ve almost finished my thesis and I hope to next move from fundamental research into proof of concepts. This is where we test our hypotheses, make up working prototypes and iterate to get feedback and develop more and more.
Medical, Computer Science, Design, Biomedical engineering, Business. Working with so many people with different backgrounds and insights means the collective work evolves into something no single person could have achieve. This is what fascinates me and it’s this challenge that motivates me daily.
When you start out on a project, you have an initial idea. It’s a bit like a seed but that seeds grows with all the different perspectives. And as your understanding evolves, the concept changes. Some things work, others don’t. Every day I’m fascinated and inspired because the concept you end up with is often turned upside down and inside out from the idea you started with. I love the fact that on the next desk to you or across the room there’s always so many talented people to ask, “how would you approach this?”
Together, we really can achieve so much more. That’s what I’ve learned at Philips. It’s really important to communicate, present and help people understand your work so you can take it to the next level.
It’s very different to university, where you work away and hand in a report. The work can be done in isolation and then you get your grades. Here, it’s very important how you communicate and champion your work. It’s such a big organization that you need your networks and you need to present your work and get backing and noise around it. It’s interesting to learn the communication part too. I’m doing the final technology presentation for our department so I’ll let you know how that goes.
