Featured Post

Do Androids cry when they deliver bad news to patients ?

How will humans - both doctors and patients - react when robots take over a large measure of responsibility for our medical care?
The GFE Robot Platform (Photo credit: Courtesy)
The GFE Robot Platform (Photo credit: Courtesy)

Just this morning, I was reading a review article by a very well known personality in the medical technology arena. In this article, the following statement was made: “But physical visits with doctors will never be replaced for important, serious matters that require face-to-face conversations”. I purposely am not including a link to the article, because it is not my intent to openly challenge this individual. On the contrary, I regularly read materials from this person, and he is unquestionably on the cutting edge of all things medical and tech.

I have previously discussed the potential of computers and artificially intelligent interfaces to mimic human beings. And I personally have no doubt that within a few decades [if not even 10 years from now], many people will not be able to distinguish between a computerized talking head versus a human, videoconferencing through a program like Skype or Google+.

The intent of the “serious matters” statement above is to support the argument that the subtleties of human interaction are beyond the scope of simulation. Firstly, I rarely if ever anymore, use the term “impossible” when speaking about the potential of silicon. Predictions within the tech realm most often tend to be wrong, unless they speak of advancements that are surprisingly fast and far beyond what we previously imagined. The sum total of research and development into all aspects of simulated humans [which I have previously called “Simulans”] already indicates that human interaction is quantifiable. In other words, tones of voice, facial expressions, quirks of speech – these can all be measured, stored, retrieved and then re-presented to a human being, who would fail to identify the non-organic state of the Simulan.

Human beings desperately hold on to the belief  that they are so individual, that their actions and behaviors cannot be predicted. But this could not be farther from the truth. For years already, marketing agencies have been able to look at readily available statistics to accurately predict how people will respond to a situation. In the recent American elections, population analysis clearly identified “problematic” regions that required more attention on the part of the campaigning party. The detail of this analysis, being able to focus down to individual streets and even homes with undecided voters, proved to me  that the days of any sense of privacy were long gone. Quite simply, human beings are in fact very predictable.

This actually should not be perceived as a negative thing. Our ability to appreciate and even predict others’ behaviors is often a sign of close friendship and intimacy. Knowing that my wife does not like most science fiction TV shows is simply a sign of the fact that I have been living with her for 23 years. In turn, she knows exactly how I like French Toast. This knowledge of each other reflects the reality that human beings do not dramatically change over time in their likes and dislikes and behaviors.

Facial expressions are definitely quantifiable. There are experts trained in identifying facial muscular activity, who are effectively “human lie detectors”. Excluding patients (or suspects) that are totally delusional or extreme psychopaths, facial expressions are a consistent indicator of a person’s mindset and intent. Even something as banal as the degree of openness of the iris of the eye, can indicate whether an individual is truly surprised or was expectant of the finding of the “murder weapon” in an investigation. Even people with the best poker faces can still be “read” by a sufficiently trained expert.

Computer vision effectively has no limitations. It can analyze the face as a whole, or individual portions of the face, or even individual pixels, in order to track changes in color and relationship to adjacent structures. A sufficiently powerful computer can do thousands of various types of analyses within moments, and then present this information to whoever requested it. It is absolutely not outside of the capability of a computer system to accurately read the emotions of a human being. And with this information, and a sufficiently detailed protocol, an artificially intelligent computer housed within a Simulan, could definitely generate an appropriate emotional response to a situation.

So, a Simulan could inform a family member about a patient’s illness or even death, and then read the response of the family member. Based on that response, the Simulan could then run through an algorithm that would generate a response with a vocal tone that is consistent with empathy. Of course, the Simulan could choose from a variety of standard texts in order to verbally respond to the family member. And once again, based on the verbal and emotional response of the family member, the Simulan could continue to run through the algorithm to consistently provide tried and tested appropriate responses.

I should also make clear that physicians rarely have any formal training in how to break bad news to patients and their friends and family. On TV, it seems as if every doctor is just inherently capable of being as empathetic as necessary. In real life, I have seen far too many doctors deliver horrible news and then walk away as if nothing had happened. Like it or not, many doctors do not have strong skills in the area of human compassion. Contrarily, a Simulan would always present a caring and concerned “face”.

I do believe in the human spirit and soul. I do believe that human beings carry a spark that is unique, and which will differentiate us from the machines for many many years to come. Nevertheless, a computer system will soon be able to emulate human behavior sufficiently well, so that everyone in need will have an attentive ear whenever necessary. Conversations with every Simulan will be recorded  and analyzed  in order to identify very human problems. If the Simulan identifies patterns of speech and tones of voice that are consistent with suicidal ideation, or violence towards others, the Simulan will be able to warn the appropriate professionals before anyone gets hurt. Whereas human physicians can easily miss a pathological dynamic [because the physicians are too tired or are not sufficiently trained], Simulans will most likely be able to pick up on subtleties that are consistent with being abused or abusing others. This kind of information will understandably be critical when assessing, say, a child who presents with recurrent abdominal pain. In these cases, Simulans will not only be helpful, but they will be far superior than humans in picking up key elements, necessary for an appropriate diagnosis and management.

Pride is not always evil, but it can definitely blind people to what is evident around them. Human beings want to believe that they can never be replaced. People have a desperate need to continue to think that their heart can never be duplicated on a circuit board. But the new reality we are living in teaches us that, at the very least, computers will soon be able to “fake it” extremely well. The obvious question now is what happens when Simulans are better at managing the emotional needs of patients, better than any human doctor. Unfortunately for most human physicians, I think the answer is obvious.

Thanks for listening

My website is at http://mtc.expert

About the Author
Dr. Nahum Kovalski received his bachelor's of science in computer science and his medical degree in Canada. He came to Israel in 1991 and married his wife of 22 years in 1992. He has 3 amazing children and has lived in Jerusalem since making Aliyah. Dr. Kovalski was with TEREM Emergency Medical Services for 21 years until June of 2014, and is now a private consultant on medicine and technology.
Related Topics
Related Posts