Healthcare is expensive. According to the World Health Organisation, one-tenth of the world’s gross domestic product (GDP) is spent on healthcare. And the US spends more than most (17.9% of GDP).

And it only going to get more costly. As demographics in developed countries shift and populations begin to age, the prevalence of chronic, non-communicable diseases is shooting up. Dementia, Parkinson’s disease, arthritis, hypertension, certain cancers, cardiovascular disease and other diseases where age is a risk factor, are all on the rise.   

Our best bet for reducing tomorrow’s healthcare costs is to prevent disease happening in the first place. To do this, we need to understand a lot more about populations. Why do Andorrans live longer than anyone else? Why do African American men have the highest incidence of prostate cancer of any ethnic group in the world?

 “We don’t know much about human populations, in general,” says Dr Andrew Kasarskis, co-director of the Icahn Institute for Genomics & Multiscale Biology. “And if we’re trying to keep a population of people healthy we need to leverage whatever data we can get, anywhere.”

Fortunately, the amount of medical data that is being created is exploding. Patient records at hospitals and GP’s surgeries are rapidly being digitised. Mountains of data are being captured in clinical trials, on medical insurance claims, on health apps, and even on social media.

One way to use this information is to provide highly personalised care. Combining a patient’s test results with data from a fitness app, such as their physical activity, heart-rate, calorie consumption and sleep patterns, produces an extremely detailed profile of their health and well-being, which can in turn be used to design treatments that are specific to their particular condition.

But the data can also be used in aggregate in order to improve treatments and healthcare processes. Data-mining studies, for example, are starting to be used by medical researchers for pharmacovigilance (drug safety).  Small datasets usually lack the power to detect adverse events, especially if they’re rare. But what if you could query the data on millions of people who have taken a particular drug?

A recent data-mining study, published in the journal PLOS ONE, analysed 16 million anonymised clinical documents to see if there is a link between proton-pump inhibitors – devices that are used to treat acid reflux – and cardiovascular disease. The study showed that using a proton-pump inhibitor increases the risk of cardiovascular disease by 16%.

As the technology required to integrate and analyse all this data becomes more widely available, the biggest challenge becomes patient privacy.

In the context of medical research, patients’ data is typically anonymised but individuals are understandably sensitive about the use of their medical data. It might reveal that they are being treated for alcoholism, drug addiction, depression or schizophrenia. It might show that they have been the victim of domestic violence or been treated for a sexually transmitted infection. Even in an age of social media and “oversharing”, few people would want to reveal these sorts of details about themselves. What if they fell into the hands of potential employers, insurance companies, marketers, vengeful ex-partners or even criminals?

NHS England’s data sharing project, Care.data, has suffered various setbacks partly because of the contentious issue of privacy. Although the data is “k-anonymised”, making it very difficult to identify individuals in the database, Ross Anderson, professor of security engineering at Cambridge University, says that it is impossible to guarantee the anonymity of patient data.

In the UK, people have been given the opportunity to weigh up the pros and cons of sharing their personal medical data. Tens of thousands of people have chosen to opt out of the Care.data scheme, presumably on privacy grounds. But millions more have not.

This would seem to imply that people are broadly comfortable with the idea of researchers using their medical data if it has been properly anonymised, but it is questionable whether they are fully aware of what the scheme involves.

Would you share your medical data with researchers? Do you believe it is possible to fully anonymise this kind of information? Share your thoughts by joining the Future Realities LinkedIn group, sponsored by Dassault Systèmes.

Sponsored by: