Should I study towards a MD/PhD in light of advances in AI/robotics/automation?

[piggyback]

I'm considering becoming a radiation oncologist. Specifically, I would like to do a BSc in Electrical Engineering plus pre-med, followed by a joint MD/PhD (EE) degree. A BSc takes four years, and MD/PhD programs take an average of 7-8 years. On top of that there is a one year internship and a residency requirement of at least five years. All in all, it would take me 18 years to get to my dream job; even if I did away with the PhD and only went for a regular MD, it would still take 14 years to become an oncologist.

My question is twofold:

(1) Will AI/robotics/automation have made most health care jobs redundant (or at least reduced their financial viability) by then, making such an investment (nearly two decades of opportunity cost) futile?

(2) Provided physicians are still needed in the future, is it at all possible to start a healthcare technology company on the side to make use of my technical knowledge while working as a doctor part-time? Would hospitals or private practices be prepared to accommodate me? Would it be financially feasible?

Thank you.

[amrosado]

I'm an MD-PhD student with the type of background you are interested in developing. Radiation oncology is going to be a dying field in the future, but that is the subject of a different conversation.

I think you want to go into MD-PhD, but I would suggest against it if you are more interested in the technology than what is best for patients. A lot of technologists are having problems conceptualizing this because they don't understand the limitations of current technology and what dealing with patients entails. Instead of trying to be a leader in this type of field you should focus more on potential problems technology can solve much better for patients by improving outcomes and decreasing costs. The biggest problem in medicine right now are the insurance companies and healthcare administration practices which are quite costly and provide little patient benefit.

1.) Technology is only going to have as big of a impact on medicine as doctors and patients allow. If you want to make an impact on healthcare focus on developing technology where you can convince doctors that outcomes and costs are better. If you look at past clinical research done, a lot of technology did not produce the beneficial outcomes perceived by the inventors. Clinicians are skeptical of technology without proper evidence suggesting its usefulness. Your technology will have to navigate this system.

2) MD-PhD gives you a lot of flexibility with doing this type of thing, but I would suggest that you focus more on helping patients than building a business. Likewise, I would suggest trying to become more involved at an academic hospital that could support your intellectual property pursuits, give you access to the patients your technology can help, and help find the resources you need for new developments. Neurosurgeons with engineering backgrounds are probably the most successful in this respect.

You have a long ways to go and the journey is not easy by any means. Good luck. Questions are welcomed.

[piggyback]

Thanks for your input. Regarding insurance companies and healthcare administration, I don't think those are problems to be solved by MDs or even PhDs but rather issues to be solved by already existing tech automation companies in coordination with legal experts. You mentioned radiation oncology is a dying field. Could you elaborate on the future of the various specialties (maybe top/bottom 3) as far as you can tell?

[amrosado]

I think radiation oncology is going to be displaced by emerging disciplines in medicine. I would pay close attention to what is happening in the field of immunotherapy since there are a lot of recent success in difficult cancers, autoimmune diseases, and infections. Recently several specialists have supported immunotherapy's potential in cancer treatment (http://www.nature.com/nature/outlook/cancer-immunotherapy/). Unfortunately, training in radiation oncology will most likely be different than the training required to perform this type of medicine. Likewise, the potential for automation/technology/AI in this particular field is quite limited.

It's important to think about why medicine might be moving away from radiological therapies. They are poorly tolerated by patients and expensive (http://www.cancer.gov/cancertopics/coping/radiation-therapy-...). It's difficult to say what will happen in the next couple of years, but radiology in general is taking a hit since hospitals are trying to limit the use of expensive and unnecessary imaging often at the expense of radiology professionals (http://www.auntminnie.com/index.aspx?sec=ser&sub=def&pag=dis...). Choosing a career path in medicine should be more about how you like interacting with patients.

[piggyback]

Can you expand on potential future medical disciplines and what specialties you think are going to be replaced?

[kerberus]

Hi! I'm an MD/PhD student, in my final PhD year (6+2, Netherlands). I'm working on computer navigation in orthopedic oncology, basically objective navigation in the OR. This, together with some small sidesteps into 3d printing and computer supported diagnostic imaging. The combination of medicine and technology is an awesome field to be in. There are so many projects that you can work on.

As Xaa has written, most doctors are not focused on technology. They will use it, understand it but especially the older ones do not embrace the possibilities automation or innovation can offer. That coupled with slow development and access to the market makes us lag behind other high-tech/high-impact industries (for example aviation).

Do not underestimate how hard it is for automation to completely take over a doctors job. In 50 years we will still need radiologist. The tools you will develop will support your and others workflows, making healthcare better. And if it replaces a certain task, others will arise, as for example radio frequency ablation has a growing role in oncology. (and is often done by a radiologist!)

Furthermore there are legal implications. Surgery will not be completely automated (fire and forget)in the next decades, simply because of this. A surgeon always has to be present. For example: the most chosen approach for robotics in orthopedics is assisted surgery, where you move the tool and the robot blocks you from making bad moves.

So, as an MD/PhD you will be a bridge between two completely different cultures. I sometimes joke that the engineers we talk to have a solution for a non-existing problem and doctors no technical solution for an existing problem. It's actually really hard to understand each others fields. So in this you are valuable!

You seem very motivated! I can only draw conclusions on my choice, but I can recommend it! If you have any questions ask away.

[kerberus]

Forgot to answer the questions directly and editing does not work:

1) No, definitely not.

2) Yes, lot's of research minded doctors/phd's already do. From producing transplant transfusion fluids to pedicle screw insertion simulators or acetabular cup reconstruction prosthesis after hemipelvectomies. During residency its almost impossible though.

[xaa]

I'm a PhD in biomedical sciences (biochem/bioinformatics), so I don't know directly about the MD side, although I work with MDs frequently.

From what I have seen, physicians are generally not tech-savvy and are fairly behind-the-curve when it comes to automation. Even many of the things that are currently possible to automate, like EMRs and some aspects of diagnosis, aren't. Since any technology that replaces what a physician does has to be approved by the FDA, it moves very slowly. I think it is a long time before large numbers of physicians are put out of work by technology. Especially since the quantity of MDs is artificially limited.

That said, radiology is probably one of the first specialties that will be automated. Already, some image recognition algorithms have been shown to outperform trained radiologists at recognizing, e.g., cancer. So, if you're purely after job security, it might not be the best specialty to choose.

But with a tech-heavy background like an EE, you have plenty of options. MD/PhD + tech background is a perfect preparation for research, if you're into that. Or you can help develop systems to automate various aspects of medicine. Although it will put people out of work, in my opinion automation in medicine is a very good thing for society because algorithms don't make mistakes (other than the inherent limitations of the algorithm), don't get tired, and you don't have to pay them, so automation should lower soaring health care costs. You can always go into a biotech firm as well.

To answer 2), yes, physicians (at many places) have good flexibility with their hours. The only thing you might have to worry about is who owns IP, especially if you work for a university health care system.

[piggyback]

Thank you for your reply. Would you be so kind to elaborate on what you do for a living as well as in what capacity you work with MDs? Frey and Osborne (2013) published a paper that indicates how likely certain professions are to computerization. You can find it here:

http://www.oxfordmartin.ox.ac.uk/downloads/academic/The_Futu....

However they didn't elaborate on the various medical specialties. Do you happen to have any information on that?

From what I hear MD/PhDs do research about 10-20% of their time, while they spend the remainder on treating patients. I'm afraid very few hospitals/practices would be willing to let me work only 3 days a week so I can work on a tech company on the side. I'd think maybe I could do some consulting but that's it; I don't think there would be time to pursue outright entrepreneurship.

[xaa]

I am a researcher at a nonprofit research institution that does both basic and clinical research. When I work with MDs (or MD/PhDs), it is usually to analyze clinical and/or genomic data that they have collected from patients. Sometimes I am roped into building web applications either for data analysis or patient questionnaires. Once in a while, I develop prototype algorithms to aid in diagnosis from e.g., histology images.

The ratio of research to clinical practice for MD(/PhD)s is basically whatever you want it to be. At my institution, it seems more like 70% research / 30% clinical on average, but that's because it's a research institution with a clinic, not a clinic that does some research. Several of the MDs I work with have dropped clinical practice altogether for full-time research as well.

Also at my institution there are researchers and/or clinicians who spend most of their time on building a business. The administration very much encourages this because they get a cut of the IP royalties. In academia, your value is basically proportional to the money you bring in, so these people are actually treated like gods.

Basically, there are a LOT of jobs out there; if you are qualified and productive, you can easily find one that suits your preferences.

(I have no hard data on what specialties are more or less likely to automate. I would expect GPs/family practitioners to be the least likely, but who knows.)

[piggyback]

That sound pretty good to me! Can you tell me about the salary range of those MD/PhDs doing research and about the cut such an institution gets? I'm asking I'm guessing the hours that your institution are sane (read: 9-5ish)? Thanks a bunch.

[xaa]

I live in an inexpensive part of the country, so salaries obviously vary based on that. $80-150K is the normal range for a PI here, but the sky's the limit if you can pull in a lot of grants/royalties.

There are no set hours. You work when and where you want. (I usually roll in around 11-noon). They don't really care how you spend your time, as long as you publish and get grants. Patents are just icing. Obviously clinical hours are in the 9-5 timeframe though.

I don't really know the normal IP cut in detail. I believe it is roughly 50% of the patent royalties, negotiable depending on how much pull you have. But if you are the actual owner of the business as well as the patent creator, then you get both the royalty cut and the business profit, whereas if you just sell the patent you get only the royalties (but obviously that is a lot less work).

EDIT: I feel compelled to point out that it's not all roses. The flip side of the freedom is that you are judged solely on your results. If you can't produce, for whatever reason, you're going to have a hard time. There's little job security. And so forth.

[djokkataja]

Do the BSc + premed and evaluate the medical thing more deeply in your third year--or graduate, work for a couple years, and then decide. Work can provide a lot of perspective that's difficult to get otherwise (assuming you're still in high school). It's tough to plan your life that far in advance; you might develop more specific interests as time goes on (plus expectations of what the industry will look like may change over the next few years too). Even if your plans change considerably, EE + premed is a solid background, so transitioning to a different field wouldn't be a huge pain.

1. Automation will have a big impact on medicine, but it will be an ongoing process, not something that happens overnight. If you're doing a joint MD/PhD(EE) program, then you'll presumably at least have some technical qualifications that would make you more appealing to a company that is working towards the automation of healthcare.

2. You might be able to start a business on the side, but not as a resident unless residencies become immensely less stressful and time-consuming. Part-time work as an oncologist may be hard to negotiate early in your medical career, but I am not a doctor (though I did investigate similar questions a few years back, and this was the impression I got from speaking with doctors and MD/PhDs). Financial feasibility... as an MD/PhD, you shouldn't have much in the way of personal debt (since the PhD covers the MD tuition as well), and I wouldn't expect you to be dipping too close to the poverty line.

Realistically, it depends a lot on whether you have a specific business idea in mind and what your drive looks like. Those aren't things that you can plan very well 20 years in advance.

[mulcher]

Yes get the MD/Ph.D With an EE degree you can build the next generation devices.

[salaaaami]

do NOT get an md degree. it is useless. in this era of corporate driven health care that is mass produced, physicians are glorified trades workers. Real impact in the tech of health care is done by engineers. Being a physician does not allow you to implement anything in the current medicolegal climate. And understanding the reality of health care delivery does not require an md degree. Witness all the lawyers pursuing suits for many of the tech advancements of yesteryear-artificial hips, slings, implants-not to mention medications.

Future development of health tech will get hamstrung by the same sort of thing except it will be driven by cyber failures and privacy violations. Physicians will remain the fall guy for any product failure. Better to get a phd in a product that has major health implications-for example, develop a sensor for blood sugar that is non invasive, develop an algorithm and an app that predicts MI and cardiac disease for an individual. etc.

We are currently importing our physicians because other countries produce them so cheaply (just like software engineers). And we pay them a fraction of what medical school tuition accumulates to.