>A technology revolution is transforming the healthcare industry, changing everything from how patients are diagnosed and treated to our battle against some of the world’s most serious diseases. It’s a revolution fuelled by new sources of healthcare data and powered by big data analytics – and it’s being pushed even further by new developments in AI. Between growing populations, ageing populations, drug-resistant microbes and pressures on staff and budgets, healthcare faces some enormous challenges. Yet with data, analytics and AI – supported by new cloud, storage and processor technologies – the industry is moving in the right direction to meet them. This revolution will change and save patients’ lives.
Its foundation is the growth of healthcare data. On the one hand, initiatives like the SAIL (Secure Anonymised Information Linkage) Databank are collecting, pooling and anonymising data, ready for research through analytics. Operating in Wales, SAIL has collected over 10 billion person-based data records over a period of 20 years, using it in projects finding links between social deprivation and high mortality rates following a hip fracture, or uncovering relationships between congenital anomaly registries and maternal medication use during pregnancy. Similarly, projects at the Wrightington, Wigan and Leigh NHS Foundation Trust are finding operational uses for their own large datasets, using them to monitor time lags between referral and treatment or ensure staffing levels meet demand during peak periods.
On the other hand, clinicians are finding ingenious ways to make use of the wealth of data collected by fitness trackers, smart watches and healthcare apps on smartphones – not to mention information being freely and publicly shared over social media. While privacy concerns won’t melt away overnight, researchers hope that, given assurances, the public will back the wider sharing of health information, particularly if it can help us fight diseases or make more informed choices about our diets, our sleep and our exercise regimes. With anonymisation and other appropriate safeguards in place, plus the legalities dealt with, there are endless applications.
From precision to prevention
Many of these harness the power of big data analytics, using these massive datasets to spot patterns or even predict outcomes based on certain factors or criteria. One large study combines genetic information with data from other studies and canSAR, the world’s largest database for cancer drug discovery, to identify pathological mutations and match them to potential drugs. Such studies are finding that, by picking out new genes involved in the development of, say, prostate cancer, they raise the chances of creating bespoke drugs to battle specific mutations. Similar studies hope to isolate the impact of diet and exercise on diabetes, so that sufferers get more motivation to make potentially transformative lifestyle changes.
Clinicians and data scientists refer to this approach as ‘precision medicine’ using analytics to find out what fuels specific variants of a disease in specific individuals, then identifying the right individual treatment path to manage or cure it. Nor is this the only way analytics is transforming healthcare. Researchers fighting antibiotic resistant superbugs hope that analytics could find answers there in the long term, and that, in the shorter term, mathematical modelling could help estimate the global impact of antibiotic resistance and make a powerful case for increased funding.
Meanwhile, new healthcare apps, like Sentimento Ltd’s My Kin, are working to help prevent illness. They do so by bringing in information from smartphones and wearable devices, including physical and social activity, sleep and environmental factors, and then using analytics to pinpoint behavioural changes that could help reduce health risks and prevent the users from developing serious conditions later.
Putting AI into practice
These approaches are only being improved by developments in AI and machine learning, as clinicians and researchers use new techniques to spot patterns faster or get a more accurate diagnosis in less time. At both MIT and the University of Pisa, smart algorithms are enabling MRI image scan comparisons that used to take up to two hours to be done in one thousandth of that time, or to cut down the time patients spend in discomfort during vital MRI screenings. Similar work is being done by Intel and the AI company, MaxQ, to analyse CT scans of stroke and head trauma patients to reduce error rates, or by Intel and the med-tech company, Novartis, to analyse thousands of images of cells during drug research and identify promising drug candidates. By augmenting manual analysis, the technology can reduce screening times from 11 hours to 31 minutes.
It’s even hoped that by combining data analytics and AI, the kind of cancer drug treatment research mentioned earlier could go on to not just target the right treatment but prevent the disease from establishing a foothold. Machine learning and deep learning techniques could spot molecular drivers or mutations early and suggest appropriate action.
These developments make heavy demands on today’s technology; whether you’re working on large datasets in memory or pulling data from disparate sources in the cloud, storage speed and processing power count. Here fast flash storage arrays and persistent memory, like Intel® Optane™ DC persistent memory, is delivering the kind of high-performance, high-capacity storage these applications need – and making it more affordable and accessible.
Much the same is happening on the processing front, where Intel has teamed up with Philips to show that Intel Xeon Scalable processors can perform deep learning inference on X-rays and CT scans without the specialist accelerator hardware usually required. Using AI in medical imaging has been challenging up to now, because the imaging data is high-resolution and multi-dimensional, and because any down-sampling to speed up the process can lead to misdiagnosis. New deep learning instructions in Intel’s 2nd Gen Xeon Scalable processors enable the CPU to handle these complex, hybrid workloads. Through this research, Intel and Philips are bringing the use of AI in medical imaging down to a lower cost.
In doing so, Intel is helping supercharge the new technology in healthcare revolution, providing the industry with the compute and storage performance it needs to transform raw data into personalised treatment plans and great patient outcomes. What’s more, it’s doing so in forms that will only grow more affordable and accessible with time. Combine that with the explosion in healthcare data and there’s potential here for something truly special. Technology might not kill the world’s superbugs or defeat cancer straight away, but it could forge major breakthroughs in these and many more of the world’s biggest healthcare challenges.