Artificial intelligence today can predict your death

By Aswin Anilkumar

Global fertility rates are declining. Fewer babies are born to women today, as compared to the 1960s. While this has helped better manage our resources, its consequence has been a rapidly ageing population – with limited resources to take care of the aged. The National Health Service in England, which offers subsidized healthcare, is a case in point.

Worsening performance in healthcare

The NHS is among most important institutions in modern England, having been described by Danny Boyle as “the institution which more than any other unites our nation”. Despite being founded on the principle that healthcare had to be comprehensive and free, the decades since its founding have seen patients incurring greater costs. Concurrently, key indicators of performance have worsened, with doctors in the NHS describing harrowing tales of overflowing wards. When contrasted with recent assertions by Prime Minister Theresa May that the NHS is “more than adequately funded”, a dire picture of hospitals stretched thin and of a government in denial, emerges.

While this may seem isolated, it is part of a growing social problem affecting countries from the UK to Canada and Japan. A viable solution may be very close – with the growing integration of artificial intelligence with medicine.

One algorithm created from this synthesis, now claims to be able to predict one’s death.

The Biology of death

The dying “process” may be stretched out over several days, and even weeks. As a patient approaches the end of his life, he may spend increasing amounts of time sleeping. The body then begins to conserve greater amounts of energy, and the patient may consume smaller quantities of food. The oxygen supply to the brain also reduces, resulting in a decreased sense of awareness. His skin turns bluish-grey, with the appearance of red splotches, termed as ‘mottling’.

The ‘apnea test’ is administered to determine the exact moment of death. This test requires a full supply of oxygen for about two minutes, following which the ventilator is turned off. Physicians then observe for signs of breathing, the absence of which pronounces the patient ‘brain dead’. Some patients also sign a Do Not Resuscitate order (DNR), which requires physicians to let patients die a natural death after they stop breathing.

Death is generally recognized to have occurred when one or both of the following have happened: The irreversible cessation of heart activity, and the irreversible cessation of brain activity. The former is termed ‘cardiopulmonary death’, and the latter, ‘brain death’. Brain death allows for organ harvesting since the body’s organs continue to function normally.

Brain death

However, brain death as a legal confirmation of death has been opposed by some, who argue that brain-dead patients could be revived. They cite the case of Jahi McMath, who because of complications ‘died’ during a routine surgery and was placed on ventilator support. Her parents resisted awarding of a death certificate (which would have meant disconnecting the ventilator) and moved to New Jersey to continue ventilator support from their home. In the years that followed, physicians reported improved circulation and possible brain activity when her body should have gradually decomposed– proving she hadn’t truly died.

Even as technology permits the unnatural prolongation of life, the definition of ‘death’ grows murkier. Though clear tests have been defined to establish clinical death, the key factor remains the lack of response to stimuli, or the absence of a consciousness. However, medical science can continue to prolong a body’s ‘life’ much past the time of clinical death. A pregnant woman declared brain-dead could still be carried to term and deliver a newborn. Or, as with Jahi, may exhibit improved brain activity years later. The definition of death then exists for operational reasons – to allow doctors to move on from a patient beyond saving him/her conventionally.

A.I can prevent unexpected deaths

A recent report has listed ‘unexpected deaths’ in hospitals, as the third leading cause of mortality in the USA, after cancer and cardiac arrests. ‘Unexpected’ deaths occur when a patient’s vitals suddenly nosedive, leaving physicians with little time to try and revive him. While it’s impossible for doctors to be able to predict this, constant supervision could lead to early detection and consequently, a better response.

Unfortunately, most hospitals today operate under extreme financial and logistical pressures, leading to high patient-doctor ratios. Subtle changes in vital readings could forewarn about unexpected deaths, but nurses would simply be too busy to monitor each patient case-by-case. This leads to figures as high as 440,000 annual deaths in the USA owing to medical negligence.

The use of artificial intelligence could change this. WAVE Clinical Platform, developed by Excel Medical, is a constantly monitoring system which gives real-time updates about the patient’s status, by combining and analyzing relevant data across electronic medical records. WAVE displays an at-a-glance warning of potential death, by monitoring subtle variations across patient vitals. WAVE can give doctors up to six hours’ warning, thus enabling procedures which could prevent further deterioration. In a study testing WAVE’s algorithm, the University of Pittsburgh Medical Centre reduced the number of deaths to zero, from a control group’s six deaths.

WAVE thus ‘predicts’ conditions which could lead to death –leading to better managed and safer hospitals than today.

A.I. in medicine

A.I. is already used in a variety of medical projects which would have been tedious otherwise. Consider human genome sequencing: The ability to reconstruct the entire genome in a matter of minutes would revolutionize medicine, for it would allow the creation of ‘customized’ medicine which would be unique to each patient. The first time the human genome was sequenced, was a mammoth effort spanning over a decade. Using artificial intelligence, however, the task has been greatly simplified to a few hours’ work.

The prescription of drugs is a similar field simplified by artificial intelligence. By feeding a database with the patient’s symptom, doctors can obtain recommendations cross-referenced against the patient’s allergies. Deep learning algorithms have been proven to perform with a 96 percent success rate at identifying tuberculosis. Artificial intelligence cannot replace human doctors yet but is fast becoming invaluable for the doctors of today.

Artificial intelligence, rather than destroy, may one day save all of humanity.

Featured Image Source: Pixabay