Interview with Mark Gooding, Chief Scientist at Mirada Medical

Medical imaging is widely regarded as one of the core components of the healthcare industry, it accounts for at least 90% of all medical data. The application of deep learning into medical imaging, whilst still fairly new, is aiming to help radiologists sift through the information. At the Deep Learning in Healthcare Summit, Mark Gooding, Chief Science and Technology Officer at Mirada Medical, will be exploring considerations that must be addressed when translating technical research into clinical products. Mirada Medical is developing advanced technology applications that help healthcare professionals use medical images more effectively and efficiently to improve cancer care. We caught up with Mark to learn more about his work as well as his hopes for the future of AI applications.

1.    Give us an overview of your role and Mirada Medical

As the Chief Science and Technology Officer at Mirada, my role is primarily to manage the research team and to keep abreast of the latest technology developments in our field. However, Mirada is a flexible company and able to offer great breadth to any role, so in addition to making cakes/tea/coffee for the team from time to time, I do enjoy working at the interface with product management in assisting to translate our research into clinical products.

2.    How did you start your work in AI and healthcare, and which one came first?

I’d say my work started in engineering, but I’ve come to AI from the healthcare direction. As someone with an engineering background, I’m very focused on solving problems for the benefit of society, and healthcare was a field in which I felt I could make a positive contribution. While AI is a fascinating field of research, which I’ve had an interest in for a long time, I would say that it only really entered my work at the point that I could see that we could apply it as a tool to solve a tangible clinical problem. In our case, the first application was auto-contouring for radiation oncology.

3.    How has the application of deep learning aided the field of radiology? And where do you think its greatest potentials are?

Over the past few years, I think the biggest change has not been in the application of DL to radiology, but in the interest of the radiology community in these techniques. While DL has shown benefit in improving computer aided detection for tasks such as mammography screening or lung nodule detection, these are areas where machine learning was already being applied. However, I believe that the excitement generated recently in AI has led to a shift in the appreciation by clinicians of what could be done in the medical field, and a renewed drive to use technology for patient benefit.

“What its greatest potential is?” is a really interesting question. If I knew with certainty where DL could offer the greatest health-economic benefit, then I suspect Mirada wouldn’t want me disclosing it in this Q&A! However, one area where I see real, and perhaps under-estimated, the benefit is in our understanding of clinical data. My experience is that the need to build curated dataset to train the systems leads to more in-depth analysis of the clinical data in ways that help the clinicians to better understand it themselves.

4.    How do you think deep learning will transform healthcare more generally in the coming years?

I think we all hope that deep learning will enable healthcare to better fulfill its purpose of care provision and that improvements in efficiency and automation of more mundane tasks will free clinicians to do the more complex and the more caring aspects of their profession.

However, as I mentioned in the question about its greatest potential – I believe that while we currently focus on the data we already have, deep learning will drive us to consider what further questions we could ask and what new data we need or want to collect. The challenge that DL introduces with respect to trade-off between safeguarding patient privacy versus developing technology for clinical benefits will only grow as we endeavour to answer more complex questions. Therefore, we could expect to see the greatest change in how we treat our personal healthcare data.

5.    How is your research transferable? Are there any other industries that can benefit from the way you’re using DL?

Wow. I’ve not actually been asked that question before. To be honest I’ve not really thought about it. I regularly ponder where else in healthcare we could apply our approach, but not other industries.

From a general perspectiveI guess many industries could and will apply DL technology in the way that we have by following an engineering approach: What is the problem? Will this technology solve it? If so, let’s apply it. Without wanting to give away the punch line, that’s really the focus of my talk.

6.    We’re excited to have you join us at the Deep Learning in Healthcare Summit in Boston, after a great presentation at the same event in London last year, what are you most excited for at this upcoming summit?

I haven’t had a chance to review the full program yet, but many of the talks in London were excellent, so I’m generally excited to hear what is going on in the field at this event. My experience has been that conferences in the USA are quite different to those in the UK culturally, and I’ve always enjoyed seeing/experiencing the difference. I’m expecting the questions to be much less reserved and look forward to some lively discussions at breaks.

Early Bird Discounted tickets for the Deep Learning in Healthcare Summit, Boston, 23 & 24 May end this Friday! Join Mark & many more other industry experts in a lively discussion about the applications of deep learning into healthcare.