Part 2 of the series was simply about listing some examples from patient’s perspective to make a point how doctors are falling. It is not to say that I expect them to be right 100% right all the time from the very first diagnosis. My point is more about capacity and willingness to help the patient beyond what is obvious. One person comes with a fever, doctors writes a prescription for paracetamol. Today at most places, you do not need a doctor for that and in many cases not even a medicine. If you go to a doctor, already with your leg twisted, with foot pointing backward, you do not need an expert to tell that it needs an operation. Also, the idea of a doctor is not just limited to cut that leg open, put the bones in line, put some screws, stitch it back, get paid and send the person home. It extends beyond that to ensure, how the person comes back to normal operations in life, to guide the patient through transition to walk, to run, to play and do more complex activities. Doctor needs to check, how can he help the patient early and provide interceptions to modify the lifestyle including recommendations of medicines, if necessary, to avoid serious injuries later on. Doctor just cannot say, “oh you know human body is so complex” on one hand and on the other hand to be so confident to say,” I did my 4 tests, all good, so you are healthy”. This to me is just not acceptable. And these are the cases where we need doctors today and in future.
A patient had a common cold and he went to doctor for advise. The doctor told him to eat cold drinks with ice, eat a lot of ice cream, sleep with full air conditioner on and so on. Patient asked excitedly, ” and then I will back normal again?”. Doctor said,” No, you idiot. This way you will develop pneumonia and I am an expert of it”.
Definition of an expert is not limited to know a subject and work on only those cases that are obvious. Today, doctors have a lot of help from machines doing variety of tests. There are experts whose only tasks to administer and interpret or read those machine produced results. The results maybe in the form of images, graphs or tests. Once “this expert” writes his perspective, no other expert would look at the original results anymore. They simply trust what the first one wrote. But are those experts ready to take that kind of responsibility. Are they sure, if they made a false positive or false negative interpretation, that all other experts would agree with them? What about if they were given the same results in random blind tests, would they come to the same conclusion? I am sure any psychologist can say that it is not guaranteed to get the same conclusion from the same results by the same doctor. Yet, no one would like to take a 2nd opinion on that result. Even if you go to take second opinion, it depends on the complex inter-relations of doctors and hospitals more than the actual skills of a person.
On the similar lines, if someone has a complex situation, doctor can say e.g., “If we get 10-20% improvement, it may already be enough”. The statement itself is fine. But, there is usually no attempt to define a baseline, process, procedures, methods and measure as how to validate that statement! How would you know, if there is 10% or 15% improvement? How frequent there needs to be an intervention to check those parameters? This is of course too much to ask, when the doctor does not even have the will to do a blood test, before prescribing multi-vitamins. And that, despite the fact that medical papers and studies are against such a prescription. It is one thing that one of your friends suggests you to take Vitamin C or magnesium or some complex because she had a friend who was in “similar” situation was helped by these. It is completely different situation, when an expert, a doctor says the same thing without doing any test.
I know from my mother that she was often told that her troubles are due to stress. But that is all there is. What can she do that would help her to get out of stress? What lifestyle changes are needed? How to even assess that she is under stress? Because it is a wide blanket, you can through all your olives and oranges on it? Even if some psychologist is involved, he will be stuck with his method without trying to look at other methods or other interdisciplinary involvement to help the patient. Why? Because all this takes too much time and effort with hard to show results. That is something almost no doctor want to get involved in.
For me, many if not all of the situations can be handled today and in near future by machines. We are in the beginning of that transformation to see machines helping patients on a daily basis for regular everyday problems. If you have a fever and cold, you do not need a human doctor to help you. You can provide your symptoms to a machine and you would get the same if not better results from the machine. In complex cases, machines are and would prove to be even better than humans. Machine can analyse a complex list of symptoms, scenarios, diseases, relevant tests and medicines much better than humans. The good thing about it is that machines can immediately setup a monitor and follow-up plan as well. The results are repetitive and without the clutter of human biases e.g. proud, greed, gender, tiredness, personal troubles, politics, compensation and so on. Another positive to that is when a machine learns something new, it does not stay with that machine so that it will get more candidates for showing better results. Rather that knowledge can quickly spread to all machines, thus helping more people with improved treatment rather than still working on 10 years old methods.
Machines are also much less effected by everyday noise that humans go through. A doctor woke up in the morning and had a fight his wife. Unless he is some Buddhist monk with 40,000 hours of meditation, he is going to have impact on his performance that day. The human doctor who more often get praised for what he did, is going to develop blind spots as he gains confidence. The doctor who is in competition with another doctor for money, for number of patients, for pride or for reputation is not going to collaborate with another doctor, even if together they could help much better and more patients. A doctor who has heard the same complain again and again, build up a system to filter the statements of new patient to align it with his own pattern. This despite his intentions to do good, but he cannot help avoid his habitual brain patterns. Humans are in general supposed to have 50% attention in present while rest of 50% is divided in future and past; doctors are no different. This is further effected by what has happened during the day. Imagine a doctor who had just 2-3 patients who just would keep on complaining about this or other pain and they are long term patients with him. The doctor is bound to have disconnect from the next patient, despite his best efforts. It is not that doctors have gone through specialized trainings to stay immune to such effects.
Many may not like this idea of machine as a doctor. That to me is only an initial resistance to change and nothing special about doctors being replaced by machines. Isn’t that was the case when thermometers and stethoscopes came in picture and old method of “pulse check” was replaced slowly? I remember that as kid, my doctor although used stethoscope, would also check my pulse by hand. Today, it does not exist in the mainstream medicine industry, but being kept alive by alternative medicines. We accepted X-rays, CT, MR and many others with time, even though it is true they do not provide full picture and the interpreter fill the gaps to his best assessment and available time. I am sure people would like a machine precision being used to do the laser surgery on their eyes or to cut a part of the brain tumor, rather than just a human hand. People seems to be completely fine with devices controlling heart beats, having prosthesis, but would suddenly come to guard, if machine will print a prescription to eat tablet for fever or allergies!. Just to be clear on the term prosthesis. We are no longer limited to passive, one for all prosthesis from old ages or from world war years. Today, we have prosthesis that are manipulative and interact with body and brain to execute complex tasks. Not many are opposing the basic machine-body collaborative prosthesis any more, but now the resistance is more towards advanced machines.
Watch around yourselves how many of your friends are using pervasive or non-pervasive gadgets that measure one or many of our bio-chemical markers. People use them to plan, execute, monitor and adopt their exercise, diet or other interventions to keep them healthy. Some are even using simple planning applications to remind them to take their medicines on time. Many of these devices are backed by applications that are not only based on general health principles, but also customize recommendations to individuals. One thing that you will notice that they won’t just give you a vague statement like “do exercise” or “do exercises for your back or neck”. But you can get specific and progressive instructions based on an individual’s progress. This to me is also good, as in one way it puts the responsibility on the person and not just on the doctor. This takes away a big load away from the doctor, who today is solely held responsible for an individual’s health. There are applications and devices to measure your sleep pattern and advise on life style changes to improve sleep patterns. And remember, this is just the infancy state of these devices. The devices and applications are improving month on month and slowly industry is coming to standardization as well.
Will artifical intelligence replace doctors? In 1997 IBM’s Deep Blue defeated Chess Master. In 2011, IBM’s watson won in Jeopardy. In 2016 (merely 5 years after) AlphaGo became Go champion. Just in an year’s time AlfaGo Zero defeated its predecessor AlphaGo 100:0 after 3 days of learning.
I must say my support for these ideas is not without bias. Many years ago, I was working with my classmates to write computer programs, while studying and designing intelligent machine systems. We were talking from speech and speaker recognition, disease identification from scanned images of tongues, detecting tumors from CT scans, to facial recognitions and so on. This included not just machines for specific tasks, but in general medical professional. We were in the phase of transition from expert systems, fuzzy logics, Bayesian statistic models and others to neural networks. We were discussing about designing programs and systems that can interact with patient, with medical record systems, with pharma and health assistants e.g. home nurses, emergency departments to provide overall care. These also included developing and utilizing other technologies e.g. mobile phones and portable test machines to keep track of various records and notify nearby people in case of special needs. Note that these discussions were happening before there was any Android or iPhone. After all those years neural networks, a.k.a. Deep learning is leading the path for the complex intelligence machines. There are still many challenges for it to solve, but just in a few years the move from being Grand chess master, to Jeopardy and to Go champion is a real progress. The computer that became the grand chess master is no match to the Go Champion. This path is looking very promising.
While we were discussing about the machine algorithms, applications and systems, our other classmates were working on bio-engineering. This was the time when Human Genome Project was just coming to its end. Cross talk among classmates kept us updated on genetic modelling, molecule development, personalized medicine and other developments. Of course, I did not become any kind of expert on these topics, but they provided really interesting parallel development of future medicine as well as healthcare system. Today, cost and speed of getting individual genome sequence is drastically low and it will continue to reduce in near future. There are developments to have machines that make specialized medicine for an individual based on his genome, bio-markers and other information to avoid giving catch-all medicines. As an engineer, I am always interested to see different ways technology can be used to provide efficient and effective service to mankind. This is the reason that despite not being a neuroscientist, I am happy to see Human Brain Project efforts in Europe and BRAIN in US or Human Connectome Project.
While scientists and engineer are working to solve ever complex problems. I am more looking forward to the implementation of more easier and less risky solutions to start building that acceptance of machines in medical systems, never seen before. To draw an analogy to industry, it needs to and in my opinion will start to threat doctors the same way, as machines threatened industry workers and then computers to a lot of labor market. Today, machines in the form of Artificial Intelligence are once again threatening the jobs of engineers, lawyers, even music composer, painters and writers. I am not sure what the medical industry thinks about it and how much an average doctor thinks about it. But, I do not see them different from any other profession that can be improved by machines. If you watched the videos, you would know that there are some doctors who are who are supporting these ideas and in fact helping to build these machines.
Of course, there are many challenges; social, political, economics, technological etc, in that path. But, I do not think there is anything that cannot be solved. I also do not think that medical world, led by machines will be 100% fool-proof. That is just a wrong expectation and mostly it comes from the fear of unknown. We need to be careful not to evaluate machines to the absolute, but to see if they can improve from where we are today. it is quite natural that we try to judge and evaluate new entrant to the absolute and immediately forget about the negatives of today. We do not need a man guarding the rail-crossing any more. This can be done by a machine with a little training of people. The same way, we won’t need a lot doctors in the future to give simple pills to patients and avoid, multiple (unwanted) visits and wait time in hospitals. That can be done by machines with a handful of really good doctors.
Having said that machines can help solve a lot of problems that we face on a regular basis in the healthcare system, I also know that there are teams working on to change the system and bring improvements. There are groups working on to bring integrated health care approach. Another groups working to promote disease prevention and building healthy lifestyles and early intervention rather than waiting in the clinics for sick patients. E.g. Medical Council of India for the first time is including Attitude, Ethics and Communications (AETCOM) as and integral part of the course. How it will work and if comes out to be successful, we need to wait to see the results. However, I think it is a step forward. I hope that those limited experts that we would need in future behind machines would really value these AETCOM. I also hope that by that time there is progress made in integrated health care system, which to me can be possible with machines much more easily than human experts.
I am not sure what kind of battles would happen between humans and machines in the future. However, I am a person who likes to smile and love to spread it around. Going to hospitals and meeting doctors is no fun for anyone. However, when one goes through experiences as I mentioned earlier, more and more, the inbuilt smile of a person gets lost in frustration, anxiety and prolonged pain. I am a solution oriented person and like to see progress. If you leave a hospital with a solution, it does brings a smile to you. If not a smile then for some it brings a relief. Does it matter if that relief or smile was the result of work done by a machine? Or there are some who would insist to meet human expert for all his repeat checkups, which probably were not needed?
Mulla Nasrudin’s young wife, recently returned from her honeymoon, was complaining to her friend about her husband’s drinking habits. “If you knew he drank, why did you marry him?” her friend asked. “I DID NOT KNOW HE DRANK,” said Nasrudin’s wife, “UNTIL ONE NIGHT HE CAME HOME SOBER.”
At the end, I am leaving you with a small list of articles. I would encourage reader to read at least the abstract, if not the full article.
- Google’s lung cancer detection AI outperforms 6 human radiologists
- Deepmind’s AI can recommend treatment for more than 50 eye diseases with 94% accuracy
- Deep Learning program identifies skin diseases better or at part with human experts
- CompCyst, a computer program would have spared surgery in more than half of the patients who underwent unnecessary resection of their cysts.