Humanity has a problem: we are living longer. Between 2000 and 2015, global average life expectancy increased by five years, to 71.4. This is the fastest pace since the 1960s, according to the World Health Organisation (WHO).
This is a problem because long life tends to be expensive. Those surviving conditions that would once have killed often need continuous care, and older people generally need more treatment for ever more complex health problems that emerge in later life. This, along with rising costs of new medicines and procedures, has caused ”healthcare inflation” in excess of that seen in other industries. Between 2011 and 2016, average US family health premiums bought through employers increased by 20 per cent, according to the consultancy PwC, while wages rose by just 11 per cent. WHO data show healthcare spending as a proportion of national economies rising by a third in the UK between 2000 and 2015, by half in Japan and by two-thirds in the US. In many countries, the result is a full-blown crisis.
“The slowest sector to respond to technology is, unfortunately, healthcare,” says Rami Qahwaji, professor of visual computing at the University of Bradford in the UK, where he helps run a government-funded digital health enterprise zone.
Despite healthcare’s resistance, Qahwaji thinks there are significant opportunities. Among the most promising is the adoption of artificial intelligence (AI), which could significantly lower the burden on doctors by doing some of their work for them. Owing to advances in machine learning, which learn by a wealth of sample data to make predictions, technologists have made significant leaps in the kinds of perceptual, pattern-recognition tasks that constitute significant parts of the job of diagnosis in radiology, pathology, dermatology and other specialisms.
For the time being, AI systems will support human clinicians rather than replace them. AI is good at carrying out very specific tasks, says Elad Walach, chief executive and founder of Aidoc Medical, an Israeli start-up that has developed a system for spotting abnormal radiology scans of the head and neck, which has been adopted by healthcare providers across Europe, Israel and the US. “But radiology diagnosis is not a narrow task,” he says. The aim is to let the software do some of the more laborious work and make better use of radiologists’ time. Aidoc’s technology supports radiologists’ decision-making by prioritising cases that appear to include abnormalities and indicates where these lie within the images, but it leaves the decisions to the clinicians. Walach says that one facility has reduced the time medics spent on scanning and diagnosis by 60 per cent, adding that other areas with potential for using AI include analysis of genetic material. In the long term, it is likely that machine learning approaches will encroach on tasks based on perception, interpreting data and prediction that make up a large portion of doctors’ jobs. This may be bad news for some doctors in training, but good news for healthcare budgets.
There could be more general efficiency benefits from the adoption of AI in triaging patient complaints. Suitably informed algorithms can direct people to the most appropriate level of healthcare for their symptoms, with research suggesting that a fifth of UK visits to general practitioners and emergency departments concern minor problems that could be treated at home. Given the savings that could be realised by freeing up such facilities, innovation charity Nesta believes that it is likely that AI systems will become common as first points of contact for healthcare systems.
The development of wearable technologies, such as smart watches and bracelets already widely used to monitor fitness, is another area of innovation showing great potential to transform healthcare. Rupert Page is a consultant neurologist and clinical lead for the Dorset Epilepsy Service in southern England, part of the country’s National Health Service. The service has issued around 80 patients with Microsoft Band wrist-worn wearables linked to their smartphones, to test the band’s ability to spot tonic clonic seizures, the most serious type. Data from the band’s accelerometer can detect 83 per cent of such seizures, nearly as high as medical-grade detectors.
Although Page says that consumer wearables are not ready for use in day-to-day healthcare, he sees future potential in a variety of conditions, including monitoring a patient’s quality of life through factors such as how far they walk each day during chemotherapy or cardiac rehabilitation. This could help follow-up services become more efficient and effective by focusing them on those who the data indicate are in the greatest need, and data could also be used to help coach the patient.
Wearables could also be used to spot problems much earlier, in some cases with the use of AI to analyse the captured data, making life better for patients and saving healthcare systems money.
Qahwaji says doctors’ work could be transformed if they have access to the data gathered by patients’ smartphones and wearable devices, but adds that such technologies can also tackle threats to wellbeing, such as loneliness and isolation. This can involve linking people together, but social and mental health issues could also involve linking individuals with software.
Woebot is a chatbot designed to coach users on cognitive behavioural therapy (CBT). Alison Darcy, founder and chief executive of Woebot Labs, says that ”task-shifting” is already an established technique in CBT, with cheaper and less-qualified staff often providing initial support. “The problem is that any human in the loop, just by definition, undermines the ability to scale,” she says. In this case, software is used for those who need less support while human clinicians focus on the more difficult and complex cases.
With little cost for each extra user, chatbots such as these can be used by those with slight symptoms, potentially preventing them from developing more serious problems in future. Such systems could also be used for behavioural health, such as helping people to stick to medication or lifestyle changes that help tackle diabetes, Darcy says.
She adds that chatbots could also widen access to mental health globally. Low and middle-income countries spend less than USD2 a year per person on mental health compared with more than USD50 in high-income countries, according to a 2015 WHO report. “Therapy in general is too expensive for most people in the world,” Dr Darcy says. “Insisting that is the only way to get help is just not going to be sustainable in the long-term.”
As such, the adoption of these technologies shows great potential to lower the burden on doctors, and make healthcare in wealthier countries more efficient, slowing or stopping healthcare cost inflation. Longer term, this could free up resources to better address mental health and wellbeing. But in poorer countries, such technologies will prove essential in allowing them to adopt standards of healthcare taken for granted elsewhere.
This article was produced by The Economist Intelligence Unit. It originally appeared on:
https://innovationmatters.economist.com
