The global healthcare system is in need of a booster. Each year, $1.8tn is spent on care that makes little or no difference to people’s wellbeing, according to the World Economic Forum. Much of this is preventable: 10 per cent of hospital fees are spent dealing with infections that patients pick up while in treatment, according to the OECD.
The availability of healthcare data makes these inefficiencies a prime use case for emerging artificial intelligence (AI), says César Pérez Ruiz, chief investment officer at Pictet Wealth Management. Predictive analysis is already helping administrators determine why diseases might spread in hospitals, while personalised medicine reduces the chances of adverse side effects by prescribing drugs based on an individual’s genetic information.
So why is so much money still wasted on ineffective healthcare? The reality is that AI is still in its infancy. “Every single algorithm today has been trained with insufficient data,” says Omar Costilla-Reyes, a senior research scientist at MIT’s Computer-Aided Programming Research Group. “This means AI is still likely to make mistakes.” Major investment in healthcare technology is likely to change this.
Every single algorithm today has been trained with insufficient data. This means AI is still likely to make mistakes.
Omar Costilla Reyes, senior research scientist, CSAIL, MIT.
The value of data: moving from reactionary to preventative medicine
At the end of 2023, the UK Biobank announced that thanks to £200m of public, private and philanthropic investment over the past two decades, it had completed the world’s largest genetic sequencing project and would release the data for use by researchers around the world. By allowing physicians to distil population-sized data sets of genetic and environmental information into clear indicators of disease, AI allows physicians to intervene before an individual needs treatment. In the US, where as much as $45bn is spent each year treating patients who pick up diseases while in hospital, this is extremely valuable.
As AI becomes more adept at identifying the signs of disease, the focus of healthcare will gradually shift from treatment to prevention. “To do this, we need to move towards an open innovation model that allows health-relevant data to be shared and linked across platforms,” says Tina Woods, a healthcare entrepreneur who worked on the British government’s landmark data integration project NHSX. “This is one of the aims of legislation going through at the moment to drive the smart data and digital economy.”
To do this, we need to move towards an open innovation model that allows health-relevant data to be shared and linked across platforms. This is one of the aims of legislation going through at the moment to drive the smart data and digital economy.
Tina Woods, healthcare entrepreneur
Capturing the whole healthcare picture
Rules and regulations around the use of people’s health data will determine the way AI shapes our healthcare systems. “Look at finance,” says Woods, citing EU laws that determined how companies can use people’s financial data, and how they opened up banking to new participants who developed different digital methods of payments, loans and credit.
In the realm of finance, nearly all of our payment information is captured by a single institution, usually a bank. In healthcare, however, 80–90 per cent of the data that determines the state of our health – based on behavioural, socioeconomic and environmental factors – exists outside of the healthcare system, according to a paper published in the American Journal of Preventive Medicine.
AI-powered healthcare: who benefits?
Tech giants are investing heavily in the potential of AI in healthcare research and are in something of a race to buy up AI healthcare start-ups. Google Research’s Health and Bioscience Unit employs more than 100 researchers, and has published over 300 peer-reviewed studies, according to a study published in Nature. Meanwhile, other technology companies have formed Research & Development partnerships with healthcare institutions to train artificial intelligence on vast anonymised data sets, identifying patterns in genetic variations and where they indicate an increased likelihood of disease.
The convergence of technology and healthcare companies has raised questions about how data is being shared. “The terms of these deals are incredibly strict,” explains Pérez Ruiz. “Healthcare companies are willing to pay for ‘AI as a service’, as well as the advisory and utilisation fees, but they’re incredibly reluctant to share the data with the technology company. This is a problem for training AI.”
Healthcare companies are willing to pay for ‘AI as a service’, as well as the advisory and utilisation fees, but they’re incredibly reluctant to share the data with the technology company. This is a problem for training AI.
César Pérez Ruiz, chief investment officer at Pictet Wealth Management
Another problem is the lack of diversity in these data sets. Without access to data sets that represent the genetic and ethnic diversity of local populations, AI has the potential to exacerbate healthcare inequalities, something the World Health Organization (WHO) describes as “dangerous” to less economically developed nations. This assumes these countries have access to AI-powered healthcare, however, which requires significant initial investment and education.
If AI is to improve the economics of healthcare, it must be understood in the context of a world in which limited access to healthcare stifles economic growth and drives international migration. Providing AI-powered telemedicine or support for overburdened physicians in poor countries will relieve economic pressure at home. At the same time, prioritising training on the utilisation of AI in areas such as social care can help support ageing populations and create new jobs for lower-skilled workers.
“A big part of AI’s value will come from yielding insights into individual health trajectories through effortless data capture – via wearables and sensors, for example – helping us move from a ‘sick-care’ model to a prevention-based economy,” says Woods. “It isn’t just AI. We’re only starting to scratch the surface, with new frontier technologies like quantum computing and systems biology expected to accelerate progress massively.”
Together, these advances promise not only to transform our healthcare systems but also to revolutionise our economies. Realising these hopes, however, will require more than just innovation: it will need collaboration between governments, regulators and companies across multiple industries around the world.
