With the increased interest in wellness on the one hand, and large volumes of patients on the other, it is possible that an AI-mediated hybrid system of healthcare can arise from India.
How can India provide proper healthcare to its population? The Indian nation is morally obliged (and the states are constitutionally obliged) to provide its citizens a minimum level of health, in addition to roti, kapda, makaan (food, clothing and housing). The government of India is attempting to bring healthcare to the masses through a variety of means, including health insurance, price-controlled drugs and medical devices, and an increased push to build hospitals in even remote areas.
In an article in The Times of India (“Transform Indian healthcare”, 24 June 2019) Devi Shetty, a cardiac surgeon, points out pertinent facts: healthcare is the largest global industry with a $8.2 trillion revenue; and according to the World Health Organization (WHO), India’s healthcare system is 112th in the world, behind Iraq and Venezuela, “mainly due to the shortage of doctors”.
He suggests problem-based learning (PBL), that is, converting the 330 or so large private/trust hospitals in India with more than 300 beds into certified teaching hospitals churning out 100 MBBS doctors each per year, with plenty of practical experience.
While that may be one approach, it is clear there is both opportunity and threat in classic business fashion. Healthcare is a lucrative business, but done poorly, it can produce galloping cost inflation with poor outcomes. This has happened in the US: healthcare costs are growing much faster than inflation, already eat up 20 per cent of gross domestic product (GDP), and the US is worse in healthcare outcomes than other rich countries.
The worst of all possible worlds, in other words, and India appear to be marching down the same slippery slope with corporate hospitals, insurance, and other symptoms. Fifty years ago, middle class people used to go to government hospitals; today, they would only consider a branded corporate hospital, which charges them an arm and a leg, covered by insurance.
There are perverse incentives aplenty, often rising from well-meaning policies that end up being counter-productive. As one example, state and central governments insist that all child-births must take place in hospitals. This is understandable because adequate hygiene and access to emergency services may have saved the lives of some infants and mothers.
However, there is a mindset here which is troublesome. It treats child-birth, which is about the most natural thing in the world, as a disease. In most cases, a qualified midwife with a set of equipment can safely deliver a baby in the mother’s home without the expense and hassle (and the possibility of infection with antibiotic-resistant bacteria) of hospitalisation. Perhaps, only pregnant women with complications should be forced to deliver in hospitals.
Furthermore, it has become a money-making opportunity for corporate hospitals. The WHO notes that the “ideal rate for caesarian sections is between 10-15%”. However, a 2018 study (“High prevalence of caesarian section births in private sector facilities”) showed that private hospitals in India had 37 per cent c-sections against 13 per cent in public hospitals (also compare to 65 per cent in Bangladesh and 90 per cent in Brazil respectively, according to British journal The Lancet).
There is good reason: the hospitals often charge three times the price for a c-section as for a normal delivery; besides, it can be scheduled at a time without affecting the doctor’s round of golf; and last but not least, pregnant women want babies born on a particular star date, and they like the prospect of not having to go through labour pains.
So far so good. But the WHO notes that “maternal deaths following caesarean sections in low- and middle-income countries are 100 times higher than in high-income countries”.
There is also the gut microbiome, which is now implicated in a whole gamut of diseases ranging from gastrointestinal diseases to obesity to cardiovascular diseases to allergies to autism to depression to Alzheimer’s disease. It turns out that a baby delivered normally gets beneficial bacteria from the birth canal; so much so that some doctors are taking a swab from the mother’s vagina to smear it on the newborn’s mouth, nose and eyes. C-section babies are far more likely to be sick later, a cost to the nation.
Therefore, there are unintended consequences to blindly following the path of Western medicine. Is there an alternative? Devi Shetty identified PBL (which sounds similar to but one step above China’s ‘barefoot doctors’) as a solution for the lack of doctors. Are there other options?
We discussed the issue of universal healthcare at the Abdul Kalam Conference on Sustainable Development at IIT Madras from 11-14 July, and an ayurvedic physician, Shine Mohan Talapully was asked the direct question: “Can traditional Indian medicine scale up to provide healthcare to a billion people?” His considered opinion was that it couldn’t, because an ayurveda consultation, which is individualised, takes up too much time; and there are just too few vaidyas.
However, there was another presentation by Krishnan Narayanan of Itihaasa research, on the impact of artificial intelligence (AI) on rural India. In particular, he presented several ways in which ecosystems needed to be built to produce real impact:
- To reach the next billion, there need to be intermediaries who can transmit AI to the bottom of the pyramid
- There need to be India-specific data sets: for example, healthcare and DNA sequencing data
- There needs to be oversight on fairness and ethics implications in AI implementations
If you put together these ideas, and also consider the trajectory of Western medicine, there are some interesting insights. One of the most revolutionary ideas in Western medicine in the recent past is the realisation that treatment needs to be individualised to the patient.
Western medicine tends to treat the symptom and the disease, in dramatic contrast to Indian traditional medicine, which looks holistically at the person (which of course is why it’s much harder to provide a course of ayurvedic treatment: it is much easier to tell the patient to pop a few pills).
This mechanistic conceit, which has its roots in the Cartesian vanities of the human body as merely a machine, also makes an assumption: that the same drug will work the same way on every individual who shows a particular symptom. That is probably not so, but even assuming it does work on 80 per cent of patients, there are the remaining 20 per cent on whom it will not work.
But it also means we are depending on a statistical artifact of the randomised controlled trials: correlation. That is, the patient took the drug and the symptoms went away. But that is not the same as causation, which means you know the cause of the disease and created a treatment that would fix the root cause. It appears that, on average, it is correlation not causation that Western medicine depends upon.
It is not clear if ayurveda understands the causation, although it would appear so. They have theories about balance in the vata-pitha-kapha triune as a cause of disease. In a way, that fits in with the gut microbiome balance theory: there are good and bad bacteria, and it is their relative balance that determines whether you are sick or not.
This is where data comes into the picture, and thus machine learning and AI. There are already Indian healthtech AI firms attempting to collect enormous amounts of data and apply AI techniques to them: qure.ai is one such, and it works as a radiologist, identifying tumours and abnormalities in CAT, PET, MRI scans.
There is a possibility that AI/machine learning (ML) systems can act as a force-multiplier for ayurvedic physicians. If the diagnostic techniques used by ayurvedic physicians can be captured through ML (which of course assumes that large data sets can be created), then the AI can act as an efficient assistant to the vaidya.
Among other things, AI has been found effective in drawing out patients through unstructured questions, run at the pace of the patient. This was true as early as the 1970s, when the relatively primitive Eliza system was able to create a comfortable environment for patients to open up. Today’s far more sophisticated AI/ML systems should be able to ask open-ended questions related to a gestalt of the patient as an individual, and use the answers for diagnosis.
With the increased interest in wellness on the one hand, and the increasingly urgent need to handle large volumes of patients on the other, it is possible that an AI-mediated hybrid system of healthcare can arise from India.
That would be a wonderfully disruptive innovation in the world’s largest business sector. It could well make a few Indian firms the next big thing in healthcare, with an unbeatable competitive advantage based on traditional Indian medicine. If we wait too long, China will win there with its own traditional medicine.