Digital health transformation - 3 main types of innovation.
The COVID 19 pandemic has certainly created the momentum needed for digital health to overcome the inertia of ‘how we do things around here’ that is so deeply ingrained in clinical practice and the fitness industry.
Graham Dudley from Global Wellness Tracking in Australia interviewed Glenn Bilby, entrepreneur, physiotherapist and human movement scientist from Qinematic in Sweden. Has asked about digital health innovation and whether we would continue to operate as analogue health providers in 5 years. Watch the ‘3 Types of digital health innovation’ video.
That prompted this article by Glenn Bilby, founding CEO of Qinematic in Sweden, creators of Moovment digital health software.
The word ‘digital’ will go away in the future. Health services will be digital without saying so, as we have already seen in so many other industries. Can you imagine a taxi driver referring to a paper map to find your destination – if you booked a taxi, they already know where you are going. They just greet you and drive, without confusion. Such will be the case with clinical visits – the basic questions, basic assessments, objective assessments – they will all be available on a dashboard before the visit. Less bias, more accountability, better decisions and more time for empathy and caring.
There are three main types of innovation for physical therapists right now:
1. Process Innovation.
How did we get a pen and paper onto fax, and now fax onto a tablet so that we can communicate online faster and more efficiently? That’s is still a 2020 discussion in some countries. Sweden has some of the deepest technology penetration in the world, in homes and schools, but not necessarily in clinics, and certainly not in physical therapy clinics or gyms. Keeping an electronic journal has been a requirement for a long time, but as they say in the world of health informatics and machine learning - garbage in, garbage out – and not much of the text entries in medical journals or training diaries over the past 20 years can be used for digital innovation. It is largely just a fat and useless paper trail. In 2019, it became possible for me to see my own medical journal online, despite the fact that I have been tracking my own health data using consumer-grade sensors for more than 10 years. As a self-proclaimed ‘busy person’ I am pleased that I can make online bookings almost everywhere now. But I still need to use card payments. Almost everything from a smoothie to a taxi ride is by QR code in many parts of Asia, and when Asia is more ‘relaxed’ about the flow of information, one cannot help but wonder how quickly they will frog leap the West in terms of digital health innovation - one the faster-growing sectors with a CAGR of 27.7% between 2019 and 2025. We haven’t yet managed to use voice recognition to record patient reporting and take notes for clinicians, in order to accelerate documentation, but it is just around the corner. Many of us would like to see AI-support assist in interpreting that information. The EU is caught up in a smörgåsbord (buffet) of debate – mostly slowed down by my healthcare colleagues, not necessarily the patients concerned, who are looking online for health information (including their own) like never before.
Surely, we want to prevent patients from repeating the same answers to multiple practitioners and avoid making them say the same thing to five different doctors and five different physios and getting 10 different answers to their problems. This process innovation is good for efficiency and quality control. It's just the natural process of playing catchup with other industries who are way ahead of healthcare. Look at Amazon and how they manage information to personalise almost everything. I do wonder what healthcare would look like if it was run by Amazon – just to stretch the imagination. Do you really think that well educated and experienced clinicians would be performing mundane tasks like taking notes and measuring things that can be done so much better by a sensor or a machine?
2. Clinical Innovation
Like Amazon, at Qinematic we are using optical sensor technology and advanced algorithms to recognise objects, and how they are moving during everyday tasks like side-bend and squat. Instead of parcels moving around in 3D space, we analyse human parts moving around.
We are discovering how and why individuals move the way they do. We currently take a lot for granted, until we have a stroke or get back pain. We can see how similar, as well as how different people are. Paying attention to outliers is just as important as seeing if people conform to norms. Not that there are many norms available in the physiotherapy literature – for lack of data and slow adoption of technological innovation. If anything, they based on studying just a handful of people, not millions of people. Thank goodness we are not responsible for predicting the weather. Even the weatherman gets it wrong with millions of data points coming in every day. To really understand anything that varies over time (like the weather, and peoples’ behaviour), you need to gather data over time, and lots of it, to see the trends.
Now is the time for all healthcare workers, not just researchers, to seriously consider how we actually use the data to innovate the way we work with individuals, as well as special groups. Clinical innovation involves reinventing the way we make decisions, the way we implement interventions, the way we monitor things that work, and identifying things that don't work. In addition to becoming more efficient and more effective, and improving access and quality, it gives us the opportunity to be creative and embrace creative destruction, based on objective data and not culture or dogma. COVID-19 has certainly created a sense of urgency needed to embrace creative destruction. Healthcare Finance reported a 4000% increase in online consultations, and that 1 in 4 Americans has a telehealth visit in the first three months of the COVID-19 pandemic, most of them by video. We are yet to see what clinical innovation comes out of this.
We must be realistic, healthcare by nature is slow. It needs to be slow. It needs to be cautious because peoples’ health is important. That is priority number one. The data needs to be protected. It needs to be interpreted correctly - although that is a matter of definition. What is the correct interpretation? Classic statistical researchers will insist on low variability before reporting good outcomes and identifying some level of inference or relationship between variables. Studies are usually ‘designed’ in an incremental way, often influenced by findings from prior studies and the methods used. This sometimes becomes a game of Chinese Whispers. A dangerous game when the authors like to reference their own previous works in an effort to rank higher on the academic charts. Despite calling themselves empirical scientists, they naturally avoid criticising their own prior work. They are only human after all. Machine learning doesn’t necessarily have an agenda. It embraces variability, pulls it apart, and recognises patterns rather than direct relationships between variables. Deep learning techniques launch us into new territories of understanding. I think healthcare workers are suspicious of the ‘black box’ nature of some machine learning techniques. Traditional ways of evaluating mechanisms of injury and cause and effect relationships don’t always sit well with machine learnings uncanny ability to predict the future, and increasing outperform human decision making. Read more about the difference between Statistics and Machine Learning here.
Our own experience with researchers has been mixed. The data scientists (who know logic) are often on the cusp of creative innovation and divergent exploration, and the health researchers (how know-how messy humans are) tend to gravitate to reductionist ideas driven by medical dogma. This seems contradictory. It is a strange feeling to have a foot in both camps – one seeking uncertainty, the other avoiding it at all costs. We've been involved in studies where the research was designed around old-fashioned ways of thinking and trying to make the data fit the clinical dogma, instead of letting the data speak for itself. Our metrics, which are accurate in pixels and recorded 30 times per second, have been compared against human visual observation, to evaluate squat angles. The assumption by health researchers is that the human is correct, biased by the fact that the researchers are the humans doing the observing. Data science researchers would never entertain the idea that any human interaction is unbiased - they even question their own involvement in creating things as dry as algorithms. In addition to poor methodology, simply downgrading high fidelity insights to low fidelity and biased perceptions, is like comparing subjective performance measures from a horseback rider to the digital dashboard of a modern motorcycle.
There seems to be a need for health practitioners, more so than engineers or other science-based professions, to hold onto prior learning that has apparently served them so well over the years, or so they would like to think. As though ‘the way they get things done’ couldn’t possibly be improved by or even replaced with technology. This is possibly the greatest impedance to the progress of digital health. A combination of hubris, naivety, bounded rationality, and quite likely lack of resource-slack to experiment with novel technologies. The inertia to change trumps any great idea, especially if it was ‘not invented here’. People around the world are in large similar and have similar health needs. However, we seem to need local approval from time-consuming and overhead heavy regulators such as the FDA, MDR, Chinese FDA…the list is long, and so it the time to market. The big problem for healthcare, as opposed to other industries, is that by the time a technology reaches the patient, it is almost obsolete. We have heard from Peter Drucker that ‘culture eats strategy for breakfast’ and healthcare certainly has its own culture, depending on where you live and the attitudes that leadership has to innovation, risk, and reward.
3. Digital Therapeutics
On the intervention side, we are seeing unprecedented adoption of digital health solutions. It is improving the lives of individuals and groups of people, locally and across borders. Consumers lead the way, by taking it upon themselves to use inventions and innovations not yet recognised by the authorities, organisations, or health practitioners. It is not uncommon for patients, with their respective problems, to know more about what’s available than medical professionals. They have a vested interest to do so. Digital interventions that have a therapeutic effect are increasingly attractive due to the ease of access provided by technology. Read the book Doctor, your patient will see you now.
Moovment is not formally delivering computer-assisted interventions without the involvement of health professionals at this time, however, we have the capability to do so. Were it not for the limitations imposed by regulatory barriers, we probably would have taken this route some time ago. We consider ourselves Augmented Intelligence in that we assist health providers to make better decisions. We currently facilitate convenient exercise prescription inside of the Moovment specialist portal, rather than have a robot do it for them. We are ready to automatically generate suggestions for corrective exercises, to coach people back to health, but we haven't released it yet. We will probably release this feature in the wellness space first, even though there is a large backlog of medical cases that could benefit now instead of waiting months, if not years, for an appointment at a clinic.
The digital therapeutic space of the future is evolving rapidly and getting organised. Physical and virtual robots will do more and more of the basic stuff. For example, they will record your symptoms and put them into a decision tree that is consistent with national clinical guidelines, or whatever preferred clinical routines you have and learn to optimise the pathways along the way. It will be a super-fast and very thorough system capable of identifying even the most obscure causes of pain or illness. Back pain is a very common unsolved problems, and people are often given advice without active treatment, and usually analgesics by default. Some back pain is self-limiting and resolves after 6 weeks, but there could also be a life-threatening aortic aneurysm masquerading as lumbago (back pain) - in need of immediate attention.
In our immediate day-to-day activities, our Moovment algorithms can record and measure the way that a 14-year-old girl moves. They may measure an 18-degree knee valgus on the right side, along with slight but significant upper body compensation. In the absence of pain, she would usually fly under the radar of a coach, physiotherapist, gym instructor, parent, and even herself - despite looking in the mirror every day. With the right corrective exercise, a 4-minute scan could prevent 4 months of surgery and rehabilitation, and 40 years of knee arthritis later in life. Sadly, without the sensitivity of the technology, she may never know she needs those corrective exercises, and that could be the end of her sporting life.
Digital health adoption means sitting on volumes of wonderful, occasionally overwhelming amounts of data. In the right hands (or minds) it may be of immediate use, but if not, it can also be used in the future, when deep learning becomes more advanced and more mainstream. To see the beauty of informatics and digital health, we need to get past our traditional resistance to change in healthcare and embrace the opportunities, with due respect for any significant risks. It requires a change in mindset from ‘delivering the known’, as we have been led to believe by traditional evidence-based practice, to ‘searching for the knowable’ in real-time data from practice-based evidence. I am afraid that if we keep thinking the way we did to get us here, and don't change the way we think to get us to the next step, then the rate of adoption in healthcare will continue to lag other industries. Consumers will continue to take their health into their own hands, and many offline health professions will be superseded with online services that offer better, faster, and higher quality solutions.
Click here to watch the webinar discussing the move from Analog to Digital Health.