In the second part of our exclusive interview, Dr. Craig Mermel, Research Lead in Pathology at Google AI, on the career advice that changed his life, and what it’s really like to work at Google and Apple

 

Craig Mermel is currently the Research Lead in Pathology at Google Health where he leads a research team focused on accelerating the application of machine learning for improved diagnosis of important human diseases, especially cancer.

Prior to joining Google, Craig worked at Apple on the Apple Watch and related health initiatives. He completed joint MD/PhD training at Harvard Medical School, where his PhD dissertation focused on developing novel statistical methods for mining the cancer genome. He conducted residency training in Clinical Pathology at Massachusetts General Hospital and is board-certified in Clinical Pathology.

 

 

In what areas do you see the next big advances in AI?

There are two aspects. The first is getting AI and other technologies that have been developed in the lab to actual clinical practice. Demonstrate that they are safe, efficient and will have the desired impact. I do think we are at a tipping point and this is happening now. Many startups and bigger players are making these capabilities available and achieving regulatory approval and early commercial adoption. It’s entirely possible that we will find that some of what’s being developed is over-hyped and will not live up to its potential, but in general I believe AI is going to have a significant impact on the way we practice medicine. It’s a super exciting time.

The second is to take advantage of these AI tools, there’s a need for more digitization which will catalyze the adoption of digital pathology. I am hoping to tip the field by making pathology easier and bring us to the same level as AI in radiology, for example, a domain that has benefitted much from AI because the field was digitized a long time ago.

How far do you think pathology is away from that future you’ve just described?

The short answer is it’s happening now. We are witnessing startups seeking regulatory clearance for specific algorithms. The FDA is also actively working on updating its guidelines. The key thing is we need to demonstrate the impact. The floodgate won’t be completely open if AI hasn’t demonstrated any benefit in terms of efficiency, cost savings, more accurate diagnoses, improved patient outcomes and so on. I don’t think we are years away from seeing key milestones reached. I think the question is how quickly we can go from these milestones to every part of medicine and healthcare.

There are reasons to be both optimistic and realistic. We need AI to fit into the workflow of clinicians and that means we have to be trained to use the technology and understand it. We need to find that balance so human beings and AI can work together and perform better than a human clinician or a machine alone. I believe Apple, Google and other big tech companies have already learned their lessons. They know they cannot deploy AI into medicine and healthcare like the way they do in other industries. It’s crucial to include both tech and clinical experts in all processes from development to deployment. I don’t think we are doing something entirely new; we are pretty far along in the journey.

Many people feel that only the best of the best get to work in big tech companies like Apple and Google. Is that true?

Certainly, people who work at these big tech companies are brilliant and amazing. But I don’t feel Apple and Google have a monopoly on employing talented people. People who work in big companies have certain strengths and they are passionate about what they do. Yet, it still takes a lot of other factors to determine if they will excel in their roles.

Do you live up to the big tech employee myth of being smart but mysterious?

I have never regarded myself as the best clinician or scientist. I am just somebody who loves to translate mathematics and engineering into healthcare. In a hospital, I am probably the more quantitative person. In Apple and Google, I am a novice in mathematics and technology, but I can bring a level of clinical expertise in being able to speak across two different worlds.

How stressful or fun it is to work in these companies?

I wouldn’t complain about my experiences at either of these companies because both of them are equally remarkable. I was able to work alongside other brilliant individuals and create things that make one feel like they can have a big impact. I think, by and large, it’s not more stressful than working in a hospital or working late hours or getting paged in the middle of the night.

But of course, sometimes it can be stressful if we are racing against time to meet a deadline. In general, I guess stress comes from anywhere and everywhere since we want to do the best work possible. I believe stress, most of the time, is self-imposed because there’s always work to be done and there’s always ways to do things better.

What do you consider your biggest achievement and conversely, failure?

In the early part of my career, I was part of a massive consortium mapping cancer genomes at scale, building tools and datasets like The Cancer Genome Atlas. After which, I started dabbling with wearables with the idea that they were going to be clinical devices. Ten years later, the FDA is giving clearance to algorithms that Apple and Fitbit developed. Now, I am involved in AI and pathology, which I also think is poised to transform the field.

I feel that with each of these things, I have found myself at the beginning of a field that has the potential to grow into something massive and impactful in the years to come. I don’t think I uniquely saw that. I am the type of person who likes to work in a team to achieve things. Seeing the things I have been even some small part of reach massive scale is the achievement I’m most proud of

I’ve had many failures in my career, but I don’t tend to dwell on them as isolated instances but try to focus on, ‘What did I learn in this particular situation that I can apply to the next opportunity?’ One of the things I learned is to be patient. Often, when I become excited about something, I want to witness something happen as a result within a year. Nonetheless, if what we are working on requires a big change in the world, things are likely to go slow. For example, cancer. I am passionate about how technology is going to improve the way we diagnose and treat cancer but if I am not patient enough to wait until the technology has been validated, things won’t work in the way we imagine and it’s likely to cause unexpected harms.

There have been times in my career when I have been too impatient, when I gave up too early on what I was working on because I felt change was never going to happen. Amara’s law says “We overestimate the impact in the short-term and underestimate the impact in the long run.” That’s something I keep re-learning in my career.

What’s the best piece of advice you ever received?

Throughout my career there were many times when I could have gone a different way. It was challenging for me to decide whether I should stay in medicine or take the opportunity to work for Apple or Google. These are major shifts with a lot of risks.

My dad was a commodity trader, and his working days were measured in terms of how much money he made. He used to tell me to make sure I did something where the value of my day was not measured by the size of my paycheck. Earning a paycheck is important but it shouldn’t encompass all of me. One of my advisors also gave me a similar piece of advice. He said the way to pick a career is do something you are so passionate about, you’d do it even if you didn’t get paid for it (but of course, you should still get paid for your work!)

What advice would you give to someone starting a career in medical AI?

I talk to a lot of doctors who are interested in AI and they tend to ask me, ‘Should I take programming classes, or should I take a PhD in machine learning?’ I always answer with a question, which is, ‘What is it about AI that most excites you?’ You can’t go into tech because you are unsatisfied with your present career in medicine. That won’t work. Knowing programming and machine learning won’t get you started on AI, but passion will. You need to enjoy AI and not use it just because it’s the coolest thing in town.

It doesn’t mean that if tech works for me; it will suit you too. It’s important to understand whether or not the opportunity interests you. Many people have gone onto medical school because the idea of being a doctor is what’s interesting to them. If you don’t enjoy tech the same way and get some fulfilment out of work, you won’t be happy. Immersing yourself in a career in technology is challenging and if you don’t enjoy what you do, you won’t be as good. Like medicine, tech is also a field that requires constant practice and training.

This is also something I’d like to tell my younger self. Follow your passion and don’t worry so much about the rest. Do that and you will find amazing opportunities you could never have planned for. To be honest, I never dreamed of doing the kind of things that I have done. But looking back, all the things I have done just make sense. I believe passion has guided me well.

Where next for AI?

I believe the next thing we need to do is to deploy AI into practice. AI will have a role in future clinical practice because it redefines the way we see, understand, and learn about diseases. It’s exciting to see we have already used AI to help us learn about cancer. AI can detect anomalies in pathology images that human beings have yet to see. I think the next big advance is going to be unlocking what these things are.

 

Part one of this wide-ranging interview can be read here

Dr. Craig Mermel will be speaking at AIMed’s virtual multi-track CME-accredited event, ‘Imaging’ on 29th and 30th June.

View the full, exciting two day agenda and book here