Why scalability and ease of data integration are key to the success of AI-powered healthcare solutions
A stumbling block for those developing cutting-edge AI-powered solutions can be a failure to fully assess how to scale applications or exchange data securely and efficiently with incumbent technology systems.
That’s particularly true of healthcare, where stand-alone solutions are a rarity, particularly those that leverage technology such as AI in their workings. If you are to offer real advances in diagnosis and treatment, it’s exceptionally important to be able to integrate your own data with the masses of clinical data that already exists and share with other systems and organisations. Interoperability is a must rather than a nice to have, and innovative new solutions must be engineered with healthcare standards front of mind.
Those that fail to fully consider future integrations of AI-powered solutions are likely to struggle with the data requirements of national health systems once they move on from the pilot stage. In England and Wales, for example, the NHS has been extremely effective in introducing mandatory data standards HL7 V2 and FHIR governing interoperability, to which any applications must adhere. Although start-up and SME-owners and creators have exciting ideas and robust business plans, they often fail to fully consider the challenges of scaling and interoperability that such standards present.
This is perhaps understandable for very early stage startups but most scaling organisations would see significant benefits if implementing a data platform strategy early on, preparing them for the requirement for data to be secure, available and compliant in order to optimise the AI capabilities of their solution.
A unified data platform such as InterSystems’ IRIS for Health will take care of the data “coal-mining” to make a new product operational, freeing up valuable time for leaders who can concentrate on driving their businesses forward. The use of IRIS for Health by digital tech company, Cognetivity, is an excellent case in point.
Cognetivity has commercialised and deployed with the UK's NHS, a highly sensitive, AI-powered test that can detect signs of dementia many years before visible signs would normally occur. Early testing is key in the face of this devastating disease, which currently afflicts 54 million people, a figure which is set to rise to 130 million by 2050. The US Alzheimer’s Association estimates early diagnosis of people with mild cognitive impairment could save more than $7 trillion in health and care costs – not to mention the positive impact that early detection can have on those affected and their loved ones who are able to better prepare for the years ahead. Cognetivity’s simple, five-minute test, which it refers to as an Integrated Cognitive Assessment (ICA), is administered via an iPad. It is totally non-invasive, and results aren't impacted by practice effect or any cultural or educational bias, which are common in existing testing methods.
The test uses human’s evolutionary ‘food or fear' response to animals to stimulate large areas of the brain in a single task. Users are asked to respond to a series of split second images, some of which contain animals with varying degrees of distinction. Speed and accuracy data of responses is then analysed via the AI engine and the user's risk of having early signs of cognitive decline flagged with a previously unattainable level of accuracy using non-invasive testing.
InterSystems IRIS for Health facilitates the integration of the Cognetivity ICA platform with the incumbent healthcare data systems, ensuring that critical information gets to the right person at the right time, a crucial element required for efficient adoption and effective decision-making.
It is a great example of how data management and integration can be key in driving forward ground-breaking new technologies, including the latest advanced AI applications, like the Cognetivity test, and assuring they can be seamlessly scaled and their far-reaching benefits brought to as many people as possible.
Author:
Duncan Allen, Sales Manager, InterSystems
You can read all insights from techUK's AI Week here