Machine Learning and Future of Medicine

[ehwhatsupdoc]

Lately I've been thinking about the possible outcomes of having an IBM Watson like machine in healthcare and what this would mean for physicians. Also how "big data" and analytics will change how physicians practice. I'm of the opinion that once electronic medical records become streamlined and accessible to different hospitals and providers, it would be relatively easy to implement a computer algorithm like IBM's Watson to play a large role in diagnostics and treatment planning. This isn't far fetched. They've already shown that a computer program had a higher success rate than a cardiologist at being able to tell if a patient was having a heart attack from reading the EKG. An algorithm with a large dynamic database that could factor in a patient's history, previous medical records, current medications, blood work results, imaging results could easily outperform a human clinician in accuracy of diagnosing and treatment planning. I think this is 10-15 years away. This is already being seen in retail, Lowes just announced they will be introducing a fleet of robots to help assist customers while they are shopping. I suspect this will creep into other fields like healthcare too. So what would this mean? Better outcomes for patients. But less demand for clinicians. I think this will first be seen in IM, EM, and FM which are more algorithmic. Robots are not as good as humans with range of motion and fluidity (not yet anyways) so you'll still need nurses and then it'll be a turf war for NP,PA, and MDs/DOs most likely the physicians will win out but their scope of practice and pay will decrease significantly, then nurses will get pushed out and NP,PAs will take their place.

---

Members don't see this ad.

 

[ehwhatsupdoc]

Also check out this thread: http://forums.studentdoctor.net/threads/computer-vision-machine-learning-end-of-radiologist.1092613/

 

[ehwhatsupdoc]

This is a demonstration of IBM Watson in collaboration with Memorial Sloan Kettering Cancer Institute and how Watson is utilized in a specific case study. Very interesting!

 

[Goro]

People have been saying this for 30 years. I remember a tech support friend of mine saying that "MDs will be just like technicians." Hasn't happened yet. Not all of Dx'ing can be boiled down to flow charting.

I'd be interested to see as to how you can program Watson to do a vaginal or prostate exam.

Lately I've been thinking about the possible outcomes of having an IBM Watson like machine in healthcare and what this would mean for physicians. Also how "big data" and analytics will change how physicians practice. I'm of the opinion that once electronic medical records become streamlined and accessible to different hospitals and providers, it would be relatively easy to implement a computer algorithm like IBM's Watson to play a large role in diagnostics and treatment planning. This isn't far fetched. They've already shown that a computer program had a higher success rate than a cardiologist at being able to tell if a patient was having a heart attack from reading the EKG. An algorithm with a large dynamic database that could factor in a patient's history, previous medical records, current medications, blood work results, imaging results could easily outperform a human clinician in accuracy of diagnosing and treatment planning. I think this is 10-15 years away. This is already being seen in retail, Lowes just announced they will be introducing a fleet of robots to help assist customers while they are shopping. I suspect this will creep into other fields like healthcare too. So what would this mean? Better outcomes for patients. But less demand for clinicians. I think this will first be seen in IM, EM, and FM which are more algorithmic. Robots are not as good as humans with range of motion and fluidity (not yet anyways) so you'll still need nurses and then it'll be a turf war for NP,PA, and MDs/DOs most likely the physicians will win out but their scope of practice and pay will decrease significantly, then nurses will get pushed out and NP,PAs will take their place.

 

[Gurby]

Seems possible but I think it's more on the order of 50+ years in the future, not 10-15.

 

[brainnurse]

Nothing can replace good old Physical Assessment. (Or rather, nothing has replaced it yet.) A recent encounter with bowel perf that went undiagnosed for hours - because the whole team of multiple specialties were relying on technology - has drilled this into my head.

Also, have you implemented the GWN in your hospital yet? Technology isn't about to creep into healthcare; it's already here.

 

[Law2Doc]

Algorithm (aka "cookbook medicine") only gets you so far. As mentioned, people have been trying to boil down medicine to a simple recipe for years, and while certain small sub sects of medicine permit this nicely, most of medicine really doesn't. Watson type machines at best will be a nice adjunct to an actual physician.

 

[beethousand]

I would opine that it is going to be a long time before a computer comes close to replacing a physician. One of the major values of a physician is their ability to serve as the "interface" or translator between the patient's history and exam and the assessment process. Even if you could come up with algorithms, it is going to be extremely hard to deal with the intricacies of the human experience and the way they experience and present with illness. For example (albeit a crude one), if a Peds doc utilized an algorithm every time a kiddo came in with a fever and neck pain without delving deeper, you would be doing a lot of unnecessary LPs.

 

[jl lin]

People have been saying this for 30 years. I remember a tech support friend of mine saying that "MDs will be just like technicians." Hasn't happened yet. Not all of Dx'ing can be boiled down to flow charting.

I'd be interested to see as to how you can program Watson to do a vaginal or prostate exam.

 

[jl lin]

Seems possible but I think it's more on the order of 50+ years in the future, not 10-15.

You need to be able to use all of your senses--beyond using diagnostics equipment. After than, you have to be able to reason beyond set algorithms. It's been my experience that people often fall outside the typical enough, and so it's pretty hard to take a true holistic approach with even the most advanced machines and artificial intelligence.

Also, there is the whole other perspective of a person helping, supporting, teaching, directing, and empathizing person to person. Sadly, empathy can be tough enough for some humans to grasp, much less a robodoc.

Although, the Orgasmatron from that Wood Allen movie is something some might consider as a start.