Similar to objects in a rear-view mirror, medical interaction with robots might be closer than it appears.
This article originated from two separate thoughts. The first was during a recent movie where the protagonist falls in love with a lady, only to realize later that she is a robot. What if the future of healthcare is to become like the unfolding of events in this movie? The second was my introduction to the term “large language models” (LLMs). Stated simply, LLM involves learning by Artificial Intelligence (AI) based on billions of parameters. This enables AI to learn grammar, and semantics from vast amounts of data. LLMs can generate human-like text, translate languages and answer questions in an informative way. It can revolutionise the way we interact with computers. Medical robots – let’s call them Dr Cyborg and Nurse Marvel – must have two major characteristics i.e. professional information and language skills. LLMs are critical for the latter.
Robotic surgery has been around for a long time. The technique is useful when the surgical site is deep-seated or when nimble surgical tasks are need. However, they are always supervised and controlled by humans, and never left alone. This article is about a purely hypothetical scenario where Dr Cyborg and Nurse Marvel will attend to patients in an unsupervised manner.
I must hand out a disclaimer that my domain knowledge is restricted to medical care. My understanding about technical aspects of robots in healthcare is limited to what one would acquire by superficially skimming across various press articles.
Outpatient consultation with Dr Cyborg
The physical aspects of Dr Cyborg could surely endear him to patients. Imagine a Hollywood star lookalike with an information base of a Nobel laureate. Combine this with completely flexible working hours (to the convenience of patients) and a capacity to work endlessly with no leave. He will be a delight for patients and employers alike. Moreover, he could be multi-lingual with gender programmability. Imagine entering a doctor’s chamber and being asked about your choice of language and whether you’d like to be evaluated by a male or a female doctor? What fun!
Behaviourally too, Dr Cyborg would be perfect. His courteous bedside manners, immune from personal mood disturbances, could be so endearing to patients whose grouse often concerns improper communication by a human doctor.
There might be a few trouble spots though. Dr Cyborg will not be able to genuinely empathise with a seriously ill patient. It might not be open to suggestions or permit modifications of symptoms. A meeting in the OPD chamber might seem like a question paper where answers once recorded cannot be modified. In an Indian context, multiple languages and dialects could also pose a problem. Dr Cyborg’s comprehension of one’s symptoms may faulter much like the days of my medical internship when I interpreted a colloquially stated symptom of “pain during conversation” to be due to some oral disease, while it actually implied dyspareunia (painful intercourse).
One of the greatest attributes of a human doctor is his discerning power. The power to differentiate serious from non-serious, to be able to identify actual disease from myriad vague symptoms or to identify functional, attention seeking behaviour while a patient keeps complaining of serious sounding complaints. That’s what good clinical judgement is all about. One of my concerns is that Dr Cyborg will find it difficult to perform this task and might go into a loop of endless (and often needless) investigations, when confronted with vague symptoms.
Nurse Marvel: how may I help you?
Let’s now imagine ward care by the untiring, gender-neutral Nurse Marvel. It too will be indefatigable (except for charging and updating time), forever pleasant and efficient. It could possibly perform ultrasound or infrared light guided venous punctures for blood samples or intravenous drug administration. It could record vitals in a jiffy. All records will be accurate and uploaded to a clinical database almost instantly. A doctor could even order investigations and prescriptions remotely, thus obviating a requirement for the nurse to accompany the doctor during rounds.
The taxing tasks will be the ones which require a bit of human touch. Literally so. Will Nurse Marvel be able to decide on the correct technique, determine the injuries and avoid the pain points so as to be appropriately gentle? Sometimes patients need to be observed even while they are bathing or using the toilet. They are often comfortable with human company. I wonder if patients will be comfortable having a robot in the bathroom while it is collecting and uploading data.
Three additional points about robotic clinical interactions bother me. Firstly, what happens if there are mass casualties during a war or a natural calamity. Will these devices work if there’s no electricity or internet? How soon can they be retrained when the work environment is architecturally deformed, such as after an earthquake? Secondly, what happens to the entire medical and nursing education infrastructure created so assiduously over decades? Will it become redundant? And what if, after a few years of robot delivered healthcare, we decide to revert back to human care (such as after a natural disaster)? Will we have teachers, students and the infrastructure to train human beings once again? Finally, who would be legally responsible if something goes wrong in medical care – the robot itself, its creator or the employing hospital?
In conclusion, I must admit that I do perhaps suffer from what my favourite author, Yuval Noah Harari, terms as “fear of irrelevance”. This, according to him, might be the worst fear of 21st century. Having conceded that, one needs to understand that human interactions during illness are extremely complex and delicate. An astute clinician is not one who has the maximum information but one who uses his discerning sixth sense most effectively. Hence, while the whole world might be embracing AI and robots in different spheres of life, one needs to tread with extreme care with medical usage of unmonitored and stand-alone technology.
Prof Hemant Madan is an Interventional Cardiologist and Programme Head, Cardiac Sciences for Narayana Health. He can be contacted dr.h.madan@gmail.com.