Much has been discussed about how the explosion of healthcare data can lead to better health insights and outcomes for more people, given the access to cheaper computing power and growth of artificial intelligence technologies to process huge amounts of data. However, the reality is that the current sources of data are fragmented and unstructured, making it difficult to truly harness the power of data – this sentiment is echoed by Dr Ian Chuang, CMO of Elsevier. The democratisation of healthcare seems promising, but the road ahead is challenging, as Dr Chuang shared.
Q. Intelligent computing, aka the use of artificial intelligence (AI) and machine learning (ML), is one of the pillars in the democratisation of healthcare, as stated in Stanford Medicine’s “2018 Health Trends Report.” What are your thoughts/predictions of the impact of AI and ML on healthcare in the short term and long term?
A. Like any innovation and new technology adoption, there will be an adoption journey. The current adoption in AI and ML leverage the computing capabilities that far exceed what a human can do. The areas with significant activity and adoption are in imaging and genomics, both areas are data dense. Genomics is rich with the molecular details. Looking for the connections between genetic code and the subsequent molecules, proteins and biological functions dictated and the corresponding human experience requires massive computing power. Similarly, analysing the digital signals of a radiology image to detect subtle changes that identify pathology requires an acuity only provided through computers. These capabilities are already augmenting the clinical decision-making at the bedside.
With regards to leveraging AI and ML to predict or detect clinical conditions based on big data aggregated across different domains is trickier. Mining large bodies of data is first and foremost heavily dependent on the four V’s of the data: volume, variety, velocity, and veracity. There continues to be limits to all four dimensions, so any AI/ML is only working with the best available data at any given point in time.
Much of the data that is being gathered today through routine care comes from multiple sources, such as electronic health records and health tracking devices. This data is often inconsistent and unstructured for computing, but is broadly available and reflects realistic care settings and comorbidities. To turn this into credible, real-world evidence that supports clinical practice, the data first need to be structured, coded and for AI to even begin. My colleagues and I contributed to a piece that talked about the importance of obtaining quality data before AI can truly transform healthcare – which explains why the impact of AI and ML on healthcare is still a consideration for the long term, as there are several hurdles the industry as a whole needs to overcome.
The risk, like any research and discovery, is we are drawing conclusions based on incomplete and imperfect dataset. As a result, any generated predictive model must be analysed in context what is understood through clinical trials and based on research principles. Advancing knowledge is incremental; hence, we must analyse and judge new knowledge, especially those uncovered by machines and not based on heuristics, to see if it makes sense or if it is even within the realm of plausibility. We will require ways for peer review of the dataset and algorithms associated with machine generated knowledge. This aspect of AI and ML will advance slowly and require judicious analysis and cross validation with existing knowledge that has been time tested.
In the short term, healthcare providers must start to focus on preparing clinical and patient data for deep data analytics with the help of ML and technology such as clinical decision support, which can help with the proper recording and structuring of clinical data to improve its quality.
Q. The democratisation of healthcare implies more knowledge, convenience and empowerment of the patient in being responsible in their own care. One of the continued challenges for the once-cloistered field of medicine will be determining how best to share in the responsibility of patient care with outside organizations and how to collaborate in ways that actually lead to improved outcomes for patients.
As a trained clinician yourself, what are your opinions on the mentality of medical paternalism versus the sharing of responsibility/knowledge with the patient to ensure the best outcomes for them?
A. As a physician, the heart of our practice is applying specialised knowledge to treat and improve the health and wellbeing of our patients. Realistically, my personal opinion is that the paternalistic model of health care hasn’t consistently led to the optimal patient care or experience. Knowledge has grown and medical training has traditionally followed the “see one, do one, teach one” approach. With this approach, the clinician is only as well trained as his teacher and will lag the knowledge as time passes.
Keeping current on the medical knowledge has been hard and is getting more challenging with the explosion of medical knowledge. With the Internet and healthcare content sources such as WebMD, the consumer now has access to meaningful information. Difficulties with access and escalating cost of medical care, along with dissatisfaction with the care experience, have empowered consumers to take more control in seeking their own information and practicing self-care.
Philosophically, patients have wanted to be more informed and be able to engage in their own care decisions, as opposed to be just being told what to do. When I am a patient, I personally would value a discussion regarding my treatment options and come to a shared decision with the treating clinician about the care of my own health. We need patients to be engaged in their own health and health maintenance. Clinicians only feel challenged if they believe their role is an entitlement and the relationship hierarchical. If healthcare is truly a calling and the calling is about serving and providing care for those who seek our care, then clinicians should evolve and align the care experience and the relationship with the patients that help them feel valued as a person: regarding their opinions and fears as important, and their preferences and choices as important considerations to the care decisions.
Q. How has the deluge of healthcare data impacted your work as a clinician?
A. The explosion of medical knowledge and digitisation of healthcare data creates challenge and angst for me as a clinician. The reality is that I was probably at my best in terms of my knowledge base the night before my licensing board examination, after medical school. Since then, it has been a constant chase to stay current – it’s humanly impossible to adopt a memory-based approach that can scale to the entire universe of medical knowledge. We are in an era of medicine when we know more and can do more to better the health of the population. Currently, clinical care research data and knowledge forms the foundation of evidence-based medicine, yet it takes roughly 17 years for only 14% of new scientific discoveries to find their way into daily practice! Knowledge will remain distant and disjointed from clinical care if it remains in the library only.
Therefore, the important skill is to stay current on broad principles and advances in medicine and to know when, how and where to seek out the latest information to complete the knowledge. The clinical imperative is no longer who memorised the most, but rather who can get to the most current and relevant evidence-based knowledge at the point of clinical action, in order to deliver the best care.
While the deluge of data and knowledge can be overwhelming and challenging to manage, we thankfully now have technology such as point-of-care access to digital references database and clinical decision support (CDS) to help break down that barrier to democratise knowledge and make it accessible when health care professionals need it. CDS tools make the clinical decision and action easier and more transparent.
Q. What potential and challenges do you see in technology companies such as Google and Amazon, which are not traditionally in healthcare, being increasingly focused on fostering innovation in healthcare? How can healthcare organisations collaborate more effectively with technology companies?
A. Innovative companies like Google and Amazon are looking at tackling healthcare for two reasons. There is a problem that still hasn’t been solved well, and there will potentially be tremendous cost savings if these big problems are solved. These large organisations have both political and market clout by which to remove some of the structural and process barriers of healthcare, just like Amazon changed some of the cost and process friction in traditional brick-and-mortar retail. Healthcare has similarities in terms of the traditional brick-and-mortar model, as being the only access point to medical care and “high process friction” in how care is received and paid for.
Yet, healthcare is not any other traditional retail or supply-consumer business model. The third-party payment system creates an interesting dynamic called “moral hazard.” So, improving healthcare in terms of the experience and the cost will require re-engineering the entire system. The challenge facing Google and Amazon is changing the system. Practically speaking, when it comes to reducing cost, someone’s cost is someone else’s revenue. So, my opinion is that change will be hard. There can be a new model for health care with better alignment of incentives and value, as well as the all-important patient experience.
Healthcare organisations can continue to resist change but resisting change will just perpetuate a legacy system that isn’t working well for clinicians and patients. As external entities take an interest in fixing the healthcare system, either clinicians can be passive and accept whatever outcome, or engage and participate to ensure that their and their patients’ perspectives are considered. Clinicians and patient consumers are the key stakeholders in healthcare. Healthcare organisations and clinicians must be open to new roles and care delivery processes in the future. I would encourage them to have an open dialogue and approach any solution from the inside out, applying human-centered design thinking. Otherwise, any new solution will always be constrained by a “box” defined by external forces and legacy thinking. The future solution will never be truly innovative.
One important criterion for success will require all stakeholders to start by changing mindsets and attitudes and not be adversarial. Payers are not trying to rip off the clinicians and patients, clinicians are not trying to milk the system, and patients are not trying to get services for free. Collectively, we can take strides to define a health care model for the future, leveraging knowledge and technology.
Dr Ian Chuang is on the judging panel of the HIMSS-Elsevier Digital Healthcare Awards 2019 and the APAC edition of the Awards will take place at the upcoming HIMSS AsiaPac19 Conference held from October 7-10 in Bangkok, Thailand. Early bird rates for conference registration will end soon, so don’t miss out! – more details here.