Robot devices cannot replace rehabilitation therapists

Image source

Despite the staggering pace of development in the field of rehabilitation robotics, therapists are in no danger of being replaced by the machines. For now, human beings are needed to direct the robotic devices in what has been appropriately named robot-assistive therapy. Until machine intelligence systems that can reliably and accurately make decisions are developed, we will need human beings with the particular expertise and competence to use the equipment. In all current robot rehabilitation facilities, the robotic device does not replace therapists, rather, the therapist creates the rehabilitation regimen using the robot to speed up the patient’s recovery.
It is, of course, impossible to predict how fast artificial intelligence will evolve within the next couple of decades. Perhaps in 15 years machine intelligence would have progressed sufficiently, and creative thought processes and decision-making skills will no longer be the exclusive domain of human beings. For the time being, the usage of robotics in rehabilitation, and more generally in most branches within the medical field, is assistive to the therapist. That does not mean that the devices are not essential in refining and facilitating the work of medical staff. They can provide deeper insights, more precise and objective data that illuminate the very principles of rehabilitation, and what actually makes a patient’s recovery successful. A greater understanding of the underlying mechanisms of the therapy will in turn reveal more information on key neurological and biomechanical factors essential in radically improving the odds for recovery.
 
Human and machine intelligence unite 

While still at a relatively embryonic stage in its development, artificial intelligence (AI) is steadily entering the workspace in medical facilities, both clinical and research-oriented. AI devices are broadly defined as intelligent systems that are acquiring the ability to learn and even think. Though, largely unfounded, the fear that smart machines will make humans obsolete in the work space (once they become more involved in the decision-making process) is still growing. It is more likely that AI devices will be complementary, augmenting the capabilities of humans, rather than replacing them altogether. Smart devices will continue to enhance our ability to address complex problems.

From an analytical standpoint, smart systems clearly have greater computational power and information-processing capacity. Think of all the applications and algorithms that extend human cognition, sorting through complexity and masses of information, using a large number of parameters to draw and extrapolate conclusions. Predictive analytics, for instance, uses complex programs and calculations to integrate data, produce analyses and evaluate optionality, considerably shortening the amount of time human researchers would have to spend crunching the numbers. Humans are however better at intuitive analyses and responses, along with some other holistic though-processes that help in dealing with problems that are plagued with uncertainty and equivocality.  Most psychologists would agree that much of human cognition is not a linear outcome of careful information processing. In fact, much of our decision-making process is informed by the subconscious, driven by more by intuition than methodical rational thought or logical inference. How many times have you heard of someone’s superior intuition, when a person relied on their gut feeling or business instinct? From a psychological viewpoint, intuitive decision-making is the realm of creativity, sensitivity and imagination, the place where past experiences and judgments are unconsciously recalled. 

The consensus among researchers claims that the partnership between smart machines and humans is and will continue to be synergistic, combining intuitive decision-making with analytic intelligence.  A recent study of cancer detection in the images of lymph node cells[1] demonstrates the potential benefits of combining inputs of smart systems and experts. In the study, when only smart devices were employed, there was an error rate of 7.5%. Pathologists had a significantly lower rate of error (at 3.5%). Interestingly, when the pathologists used to the AI systems to refine their decision-making process there was an 85% reduction in the error rate (down to a mere 0.5%).[2]
 
Health economics

In addition to beneficial synergies that can lead to better healthcare results, it is important to consider the bigger picture – health economics and the issue of funding. Introducing more robots to rehabilitation clinics is not only beneficial, it is becoming a necessity. Stroke rates are snowballing globally. In 2010, 33 million people were living as stroke survivors.[3] If current trends continue, by 2030 there will be 70 million, and 4 out of 5 survivors will leave the hospitals with limited function and long-term disability[4]. The burden of stroke from an economic standpoint is among the highest of all neurological conditions. In the first year following a stroke, the mean total direct health care cost per stroke survivor in Germany is about U.S.$ 21,500 (divided between inpatient/outpatient rehabilitation - 37%, and medical care plus services - 54%); on average lifetime costs are 3.6 times higher than rehabilitation costs within the first year.[5] Indirect costs (like loss of productivity) slightly surpass the direct costs. In addition to the growing number of stroke survivors worldwide, there is a shortage of physiotherapists. In some countries, like Germany and France, there are considerably more job vacancies than candidates as students flock to jobs that are less physically demanding and offer higher paychecks. The robotic devices will help address this imbalance between the inadequate supply of therapists and rising patient demand.

The fear of therapists dreading the thought of being replaced by a machine in the not so distant future is unjustified. A combination of smart technologies and therapists’ input will help the healthcare sector cope with future challenges.

Footnotes:

1. Deep learning for identifying metastatic breast cancer, Wang, Khosla, Gargeya, Irshad, & Beck, 2016
2. Deep learning for identifying metastatic breast cancer, Wang, Khosla, Gargeya, Irshad, & Beck, 2016 https://scholar.google.bg/scholar?q=Wang,+Khosla,+Gargeya,+Irshad,+%26+Beck,+2016&hl=en&as_sdt=0&as_vis=1&oi=scholart
3. Global and regional burden of stroke during 1990–2010: findings from the Global Burden of Disease Study 2010
4. Innovations Influencing Physical Medicine and Rehabilitation Robotic and Sensor Technology for Upper Limb Rehabilitation
5. ​PCV55 – The economic burden of stroke in Germany: a systematic review, J.A. Düvel, O. Damm, W. Greiner