Artificial intelligence and up-to-date technology in the clinical neurosciences
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.263253
Source of Support: None, Conflict of Interest: None
Medical journals, lay newspapers and the social media are full of the prospects of improvement in the care of patients using artificial intelligence (AI) and advanced technology based on it.
The term artificial intelligence (AI) has emerged from the work of John McCarthy in the 1950s, when he discussed machines that replicated human skills and could solve problems.
Whilst there is no denying that instruments based on these techniques have led to notable improvement and accuracy in our care of patients, it is important to sound a cautionary note. A recent cartoon depicted the dilemma. Two scientists were discussing whether there was anxiety on the increase in the use of artificial intelligence. The senior scientist responded: 'No, but I am concerned about the decrease in real intelligence.'
This is not misplaced apprehension. We are all too familiar with the need felt by youngsters today to use the calculator when asked to determine 10% of 1500 and the use of a mobile telephone to contact someone sitting in an adjacent room.
Documented instances of indiscriminate use of improved techniques are not new. Dr. Joseph-Ignace Guillotin (1738-1814) invented 'a machine that beheads painlessly'. Guillotin, a physician, was, in fact, opposed to the death penalty but had advocated the machine, later named after him, as the axe or sword, hitherto used for execution, did not always cause immediate death and left the condemned person in agony. During and after the reign of terror that followed the French Revolution, he was to regret that the machine was named after him. The shame prompted the family to change its name after his death.
A recent issue of the Journal of the American Medical Association (JAMA) carries a paper on artificial intelligence that poses, in its title, the question 'Will the value match the hype?' The author points out that AI is now being viewed as a magic bullet. (You will recall that this was the term used by German Nobel laureate, Paul Ehrlich, in 1900, when he conceived the possibility of killing disease producing microbes without harming the body itself. He had named the hypothetical agent for doing so as Zauberkugel, the magic bullet.) Dr. Emanuel points out that already 80% of health costs in America are from tests, procedures, pharmaceuticals, and other treatments and that massive amounts of venture capital ($8.1 billion in 2018) are pouring into digital start-ups on the premise that health care is ripe for disruption, that AI is the tool to do it, and that the winning companies will reap untold profits.
The neophyte in the clinical neurosciences may well be prompted to paraphrase Shakespeare and ask, 'To use or not to use, that is the question.'
Our 'teacher generation' (a term favoured by Dr. M. Gazi Yasargil) developed their skills over years. The localization of intracranial tumours purely on clinical findings (Macewen, Gowers, Horsley, Godlee) gave way to the use of the newly discovered Roentgen rays. Up to the 1960s, Drs. Jacob Chandy, B. Ramamurthi, R. G. Ginde, Homi Dastur, Gajendra Sinh and other Indian pioneers drew valid conclusions from plain X-ray films of the skull and spine; the ventriculogram (using air or methyl phendylate); and, when the intracranial pressure was not high, the pneumo-encephalogram. Cerebral angiography, through percutaneous punctures of the carotid or vertebral artery, followed. Some departments in our country were blessed with the availability of an electro-encephalo-graphy unit.
The use of serial changer and femoral catheter angiography followed. Selective and superselective angiography, computerized tomography (CT) and magnetic resonance (MR) scans are of more recent origin.
Despite the availability of these advances, when we compare our current operative results with those of the pioneers, we are painfully surprised by the fact that our patients are not all that much better off. In the 1880s, Victor Horsley was excising epilepsy producing scars in the brain and a glioma from the cortex responsible for movements of the left upper limb without causing any neurological deficit. William Roentgen discovered the X-ray in 1895.
Used to the CT and MR scans as our students and residents are, you will find them at a loss were you to ask them to make a clinical localization of an intracranial tumour or draw conclusions from plain X-ray films, a ventriculogram, or a simple angiogram of the brain. Should such a student or resident be forced, by reasons beyond his control, to work in a rural or district hospital, he may be overcome by panic and paralysis when he is faced with seriously ill patients.
Advanced centers now boast of robots in operation theatres, smart wireless devices implanted into the brain to treat a variety of disorders, electronic neural engineering to manipulate brain signals, cancer nanotechnology and the use of stem cells and gene manipulations. In most such centers, neuro-navigation is old hat – tritely familiar and even old fashioned, but I must confess that I was dismayed when a senior resident in one of our national institutes told me that he needed neuro-navigation to insert a ventricular catheter to drain cerebrospinal fluid.
The availability of advanced gadgetry does encourage laziness of thought, reduced creativity and paralysis of innovation to overcome local deficiencies in equipment and funding.
The decline of humanism may be another consequence. Readers were stunned by a newspaper headline on 9 March 2019: 'California man learns he's dying from doctor on robot video'. Ernest Quintana's family was devastated when a robot machine tolled into his room in the intensive care unit and a doctor told the 78-year old patient by a video call that he would probably die within days.
The implementation of machines incorporating AI will involve huge costs and learning curves, with risks to patients till expertise has been gained. There is also the built-in danger of obsolescence. The need for upgrades and the dependence on the manufacturer for operation, maintenance and improvement will also have to be factored in.
Program failures can be catastrophic, as problems created by robots and other machines may be more difficult to control than those from human error.
Several parts of our country lack stable, uninterrupted power supply. When the electric supply is cut off, the machines will be paralyzed.
Studies by experts in imaging provide data that induce sobering thoughts.
Profits from artificial intelligence in medical imaging are estimate to reach US$2 billion by 2023.
We already have radiology departments in some Indian hospitals and clinics, equipped with CT and MR scanners, but lacking radiologists. Images are sent to experts in distant centers for evaluation and reporting. Nosey parkers are told that AI and telemedicine have filled in the deficiency. Worse, we are told that an added advantage is that radiologists in such institutions are freed to perform interventional procedures which, incidentally, yield them much greater incomes.
Veteran radiologists avow that to obtain the maximum information from a test, it is essential to obtain a good history – preferably from the patient or from the referring physician and learn the findings on clinical examination. The clinical diagnosis that follows enables planning of the test and a search for the lesion. Incidental findings will be noted as such and not hamper identification of the lesion producing the patient's disabilities. How is someone sitting hundreds of kilometers away to make a clinical diagnosis or determine appropriate views and sequences?
A joint draft statement issued by the American College of Radiology, European Society of Radiology, Radiology Society of North America, Society for Imaging Informatics in Medicine, European Society of Medical Imaging Informatics, Canadian Association of Radiologists, and American Association of Physicists in Medicine concludes thus:
Radiologists are learning about ethical AI at the same time they are inventing and using it. Technological changes in AI, and society's response to them, are evolving at a speed and scope which are hard to grasp, let alone manage. Our understanding of ethical concerns and our appropriate response to them shift constantly. AI will conceivably change every part of radiology to some degree. To do best by our patients and our communities, we have a moral obligation to consider purposefully the ethics of how we use and appreciate data, how we build and operate decision-making machines, and how we conduct our business.
Albert Einstein was more succinct. 'It has become appallingly obvious that our technology has exceeded our humanity.' A recent example has been the setting up of several centers in India, treating a variety of disorders of the nervous system and the consequences of injuries that have crushed the spinal cord, using stem cells.
We can best answer the question posed by the neophyte above thus: 'Use AI and technology wisely and incorporate safeguards.'