Investing in new digital technology to drive healthcare transformation is big business these days and much of this effort is being directed toward artificial intelligence (AI) platforms. One industry analysis suggests a 40% annual global growth rate in AI investment through to 2021 and the number of companies offering healthcare-specific applications is growing by the day.
However, even with these impressive predictions, you can’t discuss digital investment without also considering how it will deliver value in various forms (cost, productivity, outcomes, etc.) and for various audiences (patients, care providers, insurers, etc.) In fact, the health and life sciences industry is undergoing a radical shift towards value-based outcomes – maximizing patient health outcomes per dollar spent.
This begs a vital question: How do we ensure that any investment in healthcare AI provides the best possible value? And that doesn’t just mean value to the organizations providing care services, but also value to taxpayers and, crucially, to patients and citizens.
This is not an easy question to answer, but there are at least some basic principles emerging which will help to better frame the problem and to guide those all-important investment decisions being made across the industry. In clinical practice, our aim is always to improve the quality of diagnosis through more effective treatment leading to better outcomes. In that same spirit, here we offer three principles to consider when assessing the value of your next AI investment in healthcare.
Diagnosis – Are we addressing the root cause or just treating symptoms?
Our first principle is perhaps the most obvious: What is the diagnosis and which business value drivers are we seeking to influence? Traditionally, this has revolved around the need to make or save money through cost optimization, process efficiency or organic growth. AI is certainly well-placed to support such initiatives. But increasingly, the problem in healthcare is one of managing constraints. In the UK, for example, emergency hospital admissions have increased by 14% since 2010, but nursing numbers have remained static and even started shrinking.
So, the question then becomes how we can free up more valuable human time to care for the sick and vulnerable without destabilizing a health system. For example, we might invest in advanced clinical decision support, using neural network techniques applied to vast quantities of clinical data, to improve accuracy and speed of diagnosis. Alternatively, we could invest in cognitive virtual agents to undertake many of the routine, predictable and transactional activities of the healthcare workforce such as booking appointments, relaying information and providing coaching or lifestyle advice.
Another way of looking at this question is this: Are we attempting to target the “submerged part of iceberg” in healthcare (as shown below), i.e. high volume/frequency interactions, where a tangible and certain improvement can be delivered in efficiency and access and quality and experience, for carers AND patients?
The Healthcare AI Iceberg: Where You Target Investments
If an AI-enabled approach to routine interactions could save even 5-10% of the time of the medical and nursing workforce, just think how that would translate into extra capacity to cope with rising demand and more complex caseloads. In asthma care, for example, a chronic disease that will soon affect more than 400 million people globally, we have estimated that almost 10% of care and welfare costs could be avoided through automated proactive communication with sufferers. This would free up the time of scarce health workers to better service unmet and growing demand.
Prescription – Therapy for the future or short-term pain relief?
The treatment needed for long-term sustainable improvement in healthcare is often tied to digital technology. Today’s healthcare digital exemplars act and think like entrepreneurs – using agile adoption, fast experimentation and continuous improvement to drive their technology vision. This is especially true when investing in technology that is still maturing and has yet to prove itself fully in terms of return on investment. So, the ability to start small and scale out over time is crucial.
In considering the principle of scalability, there are some key questions to ask: How long will it take to deploy the AI solution into a production environment and what resources are needed to get it up and running? Will the solution offer self-learning, allowing it to grow with the business and improve over time? How flexible is it in terms of deployment – is it highly specific to a single care process or disease state, or is it portable enough to integrate with any number of local processes and multiple use cases? And will it offer the opportunity to replicate or remove whole end-to-end processes, including dynamic fulfilment of requests for users, across hundreds or potentially thousands of interactions simultaneously?
If the answer to any of these questions is no, then the ability of the solution to scale with the healthcare system is likely to be compromised, as all of these elements are required in order to provide a solution that can sustain in healthcare over time. Aspirin or ibuprofen can provide temporarily relief for a badly sprained ankle, but the pain will return without a comprehensive pain management regimen; the same idea holds true when planning scalability for healthcare AI.
Outcome – Will the patient not just survive but continue to thrive?
Not all technology investments will target hard financial metrics as part of their value case. Many electronic health records interoperability programs, for example, are considered necessary to drive quality and safety improvements. Our third guiding principle is therefore to consider the non-financial impact of investing in AI.
This is a broad principle and will differ with each organization’s goals and priorities, but wherever they are starting from, one key question should always be asked: How does this impact the patient? The importance of user experience in healthcare has gained momentum in recent years, so the acceptability and usefulness to the patient must be paramount. Even professional-to-professional interactions, facilitated by AI (e.g. diagnostic support, predictive bed planning and nurse rostering) will all have implications for patients that should be viewed as a positive improvement in their health journey. More crucially, where patients and citizens are exposed directly to the AI solution, such as via a virtual cognitive agent, then the experience needs to be at least as good as a similar human interaction, otherwise it will be roundly rejected.
However, user experience is only one area of non-financial impact to consider. Others might include the ability to leverage the AI solution as a catalyst for business transformation, perhaps by creating new opportunities to serve the population and to provide innovation in delivery. Doing the same things you’ve always done slightly faster or cheaper is one thing, but to re-envision a whole service built around an entirely different perspective of healthcare’s fundamental purpose is something quite different.
More broadly, we can judge any investment in AI by the wider value which might accrue to health systems. Does it offer advanced analytics to help us understand what works and how to improve through deep customer insight (e.g. sentiment analysis)? Can we use it to cross-sell or promote other products or messages that would benefit our communities (e.g. patient self-activation and preventive care)? And does the investment represent real market differentiation and reinforce our brand identities as caring, supportive bodies that have the best interests of a healthy, thriving population at heart?
Conclusion: A healthy prognosis through investment for the future
Healthcare as a whole has never shied away from taking advantage of technology for the benefit of clinical or patient care if the business case is strong and the goals are sound. Investing in AI adds a slightly different lens to the industry’s view of tech, because of its potential to be truly groundbreaking in its impact on health outcomes and hospital operations. Embracing the three principles we’ve outlined here when we weigh up the returns on our future AI investments will go some way to ensuring that we succeed in delivering the best value possible in healthcare.