AI and the Rise of Value-Based Healthcare

A few years ago, the 158-year-old insurance company John Hancock raised eyebrows when it announced it would only sell interactive policies – where data from the insured party is used to influence their policy. In this case, it uses data from fitness trackers, a trend that has also developed among some local market players. You’d likely also know of other examples, such as vehicle trackers that help reward policyholders for driving well or infrequently.

Insurers are switching to interactive policies because these enable them to provide better-personalised service and manage risks at a more granular level. And it may be the key to unlocking much more affordable healthcare, says Paul Saunders, Product Manager for Data Analytics at Altron HealthTech. 

“We all know that healthcare is broken. There’s a huge amount of waste, and medical inflation is just exorbitant – it constantly outpaces real inflation by a significant margin. This is mainly because the current system is built on a fee-for-service model. The more you serve, the more you get paid. That’s been the paradigm in healthcare for the past 40 years, but it’s led to massive waste and price increases. There’s this realisation in the healthcare industry that things need to change.” 

That change leans towards value-based healthcare, which is based on patient outcomes and not on the number of services delivered. It includes preventative therapies and actively using patient data to make the right long-term decisions.

AI’s alchemy

Artificial intelligence helps make value-based healthcare possible by developing our understanding of data. Its many incarnations are finding different roles to play in the analytically intensive world of health services, Saunders explains. “AI describes the capability of a computer to perform tasks that we usually associate with intelligence in a human being. That would cover everything from your typical rule-based systems all the way through to the later advancements in machine learning and deep learning, and there are examples of most types of AI in healthcare.” Healthcare was an early target for 21st century AI. Watson, IBM’s precursor to the current cloud-powered intelligence, wanted to help make medical information and analysis more available to doctors. But the more recent generation of AI has opened many more doors. 

Saunders recounts six major areas where we can see AI invigorate healthcare. Self-care prevention and wellness speak to the trackers and interactive policies mentioned earlier. Tied to this is predictive diagnosis – using machine intelligence to predict if a problem is developing. Image analysis of x-rays or moles are typical examples, and this is a particularly attractive field for start-ups.

AI is also playing a greater role in triage, where practitioners must allocate and prioritise limited resources among many different patients, usually in emergency conditions. Clinical decision support is where AI delves into medical knowledge and answers a physician’s queries. The machine can read all those journals, so doctors don’t have to. Also saving time for doctors is care delivery. AI can handle specific healthcare enquiries directly with the patient or act as a virtual assistant for the medical staff, taking notes and dealing with administrative tasks. AI is also boosting chronic care by being the doctor on the patient’s shoulder once they leave the examination room, likely using Internet of things and smart devices.

These examples surface across the gamut of AI technologies, from rules-based decision engines to image analysis to deep learning, incorporating chatbots, voice interfaces and more exotic means of interaction. That there are so many use cases speaks volumes of AI’s potential in healthcare.

It’s in the data 

All six of these areas also point to the emergence of value-driven healthcare. They can support prevention, more efficient processes, and greater patient involvement. The practice of billing a patient for a barrage of tests, regardless of the outcomes, does not sit comfortably with these concepts. Yet talking about the mountain is one thing. Climbing it quite another, and in this sense, the healthcare world has a steep road ahead. Critically, nothing involving AI is successful unless it can tap a healthy and reliable pool of data. 

“We’ve come a long way in terms of processing power and sophisticated algorithms,” says Saunders. “But all of these things need large volumes of clean, high-quality data. That’s where I think one of the biggest stumbling blocks is. To enable this transformation through AI means solving the data silo challenge, but it has been notoriously hard to do so.”

Several barriers underscore the problem, and some trends are working to address them. The biggest challenges are the lack of universal patient records, the lack of digital maturity in healthcare organisations, and a shortage of appropriate information management or communication standards. 

Communications recently received a boost through the acceptance of the FHIR (Fast Healthcare Interoperability Resources) global specification. FHIR (pronounced ‘fire’) aims to standardise the sharing of healthcare information. At the same time, the Protection of Personal Information Act (POPIA) is a useful framework for managing patient information. Helping healthcare practices become more digitally  competent is a challenge, particularly among smaller companies that cannot afford the investments made by big medical players. Fortunately, scalable cloud platforms are driving cheaper yet more ubiquitous healthcare possibilities, and there is growing interest among healthtech companies to support medical SMEs as well as enterprises. 

Security in healthcare is seeing a similar trend. That leaves universal healthcare records, which may be the most daunting. But Saunders sees much progress in this regard, especially as the market grows more comfortable with value-based healthcare. Several such solutions are currently pushing for attention and market share, but he anticipates that integration and normalisation will lead to two types of records. “I foresee two major healthcare records. There will be a primary healthcare record governed around the GP and the value-healthcare team. There will also be a healthcare record that’s going to be in the hospital space and shared amongst the administrators as well. Between them we will have the data that will drive AI solutions.”

When healthcare insurers started changing to predictive, customer-centric policies, they began paving the way for a future market built on data and artificial intelligence. If the medical world can overcome its bad habits around patient information, it will be the start of a remarkable transformation. 

Originally published on ITWeb, 04 August 2020.