Artificial Intelligence has been around since 1956 but only recently have its applications to healthcare been realized. AI is the clichéd acronym that science fiction readers would associate with I, Robot or R2D2 from Star Wars. But to computer scientists, AI means the display of human-like intelligence in methods that employ reasoning, discovering meaning, generalization or learning from past experiences. It is well known that computers can be programmed to perform very complex tasks such as discovering proofs of mathematical theorems or playing chess with great accuracy and efficiency. But can they replace human doctors and nurses in patient care?
What is Artificial Intelligence?
AI is being used for automation of medical diagnosis, mechanized analysis of clinical testing, detection and screening of diseases, attachable and wearable health monitors, devices and equipment for patient management and prevention and diagnosis of healthcare challenges. The major advantages of AI, often cited, include an increased doctor to patient ratio, increased efficiency, a higher quality of care and greater reach of medical services to rural hinterlands. Even the lack of comprehensive, representative, interoperable and clean data is proposed to be corrected by the imposition of the Electronic Health Records Standards developed by the Ministry of Health and Family Welfare in 2016.
A vast uneducated population
But to the vast majority of India’s mottled melting pot of a society, densely populous and steeply unequal, the robotic interface presents an imposing obstacle that is difficult to leap across. With a literacy rate of 74.04%, a significant proportion of whom can barely read, write and do basic arithmetic, most Indians who show up at government hospitals would find it far easier to speak to a doctor first hand. It is no mystery that the young are more likely to converse with artificial interfaces then those over 45 years of age. However, those with the greatest risk factors for disease are often over 45, some may be old and infirm and quite a few may have poor vision or movement. User interfaces that are designed for a general audience fail to address concerns of enfeebled users. Literacy rates for women and those from disadvantaged communities are even lower and AI, in its present form, does nothing to increase access to healthcare for them.
Lack of Empathy and Personality
Imagine that you have been given a primary diagnosis that says that you will succumb to a heart condition within the next 5 years based on your present patterns of food and lifestyle habits, history of disease and analysis of genetic factors. Would you rather have a person talk to you to answer the immediately churning questions that are brought to your mind? Will a mere conveyance of information address the emotions this news brings and its implications for your future, your relationships, your work, friends and loved ones? AI is, after all, a product of man-made algorithms that sift data through predictive analysis. A program will only be as good as its creator. Machines do not possess the ability to emote with a real human. People are born to be social animals and they need other humans to bring them that sense of concern, empathy and affection that keeps them going beyond what any machine with limited information can predict. Even if you know the names of every chemical compound inside a person’s body, every DNA arrangement pattern, every neural pathway and centers of electromagnetic activity inside a human brain, random factors and dynamic environmental interactions preclude the possibility of any certain result.
Many doctors are apprehensive about the ability of technology to capture subtle cues such as a look, a gesture, a painful motion of the hand or a limp that may be suggestive or indicative of an underlying condition. Can people really trust a robot to decipher complex human behavior when the science to back that information is incomplete? This is not to say that statistics do not help decision making when analyzing a sample in pathology or forensics. In fact, technology is a great enabler for accurate therapeutic treatments via precision diagnostics.
But one cannot expect a chatbot to replicate the functions of a human doctor with years of experience and training under her belt. Moreover, continuous advances in knowledge and new research require constant updations to databases and application software. Chatbots can at most serve the needs of urgency and basic generic first aid as a first point of contact when lines are busy. However, they cannot pick up on subtle clues during an emergency to differentiate between a life threatening situation and an ordinary merely inquisitive query. This is one of the prime reasons why medical first responders prefer a human answering a helpline instead of an IVR.
Access to open medical data sets and associated privacy concerns have the potential to erode the sacred doctor-patient relationship. Confidentiality is a persistent need in this provocatively open society of the internet age. Social mores, mutual respect and the very real threat of misuse and manipulation prompt deep concerns for the maintenance of privacy in electronic records.
Perhaps, 20 or 30 years into the future, primary care will be pervasively handled by an AI manifestation. Various studies have shown that Artificial Intelligence has been found lacking owing to its impersonal nature, its tendency to convey sensitive information in an un-empathetic manner and its failure to pick up on visual cues from physical examination. Is this the future that we really want? The question lies open to the reader.