Vinod Khosla, the feisty VC and legendary co-founder of Sun Microsystems, has released his predictions for the future of healthcare. Below is a summary of his forecast.
Healthcare today is broken. It’s the result of approaching medicine according to practice and tradition, rather than real science primed by objective data, and unencumbered with the conflicts of interest that lead to suboptimal results. Doctors today are doing the best they can given the current system, but we should embrace the new opportunities ahead of us.
Technology will reinvent healthcare as we know it.
It is inevitable that, in the future, the majority of physicians’ diagnostic, prescription and administrative work, which over time may approach 80-percent of total doctor time, will be replaced by smart hardware and software. Healthcare will become more scientific and more consistent, delivering better-quality care with inexpensive data-gathering techniques, continual monitoring, more rigorous science and more available and ubiquitous information leading to personalized patient insight. Many new findings will be outside the reach of most physicians because of the volume of data and the unique holistic insights that data will provide about a patient’s very complex condition. Hundreds of thousands or even millions of data points may go into diagnosing a condition and monitoring the progress of a therapy or prescription, well beyond the capability of any human to adequately consider.
This evolution from an entirely human-based healthcare system to an increasingly automated system will take time, and there are many ways in which it can happen. Today’s traditional approaches will get better as new approaches, and even new medicine, will be invented. The remaining 20-percent of physicians’ work will be AMPLIFIED, making them even more effective. Doctors will be able to operate at substantially improved levels of expertise in multiple domains, and they also will be able to handle many more patients. The primary care physician and maybe even the nurse practitioner may be able to operate at the level of six specialists handling six areas of care for one patient with multiple comorbidities in a more coordinated and comprehensive manner without inter-specialist conflicts. This transition will affect each group of actors in the current system differently. Some constituencies will be affected favorably in some dimensions and worse in others, but the net benefit will be substantially positive for society and individual patients but it is likely that a focus on science, data, and personalization will lead to plenty of unintended benefits that we cannot anticipate today.
Nurses will be made much more capable by technology, often replacing the functions only doctors perform today. New medical insights, including ones we cannot yet envision, will be commonplace, and the practices we follow will be substantially better validated by more rigorous scientific methods.
My statements are not forecasts that the hospital burn unit or emergency department will run without any humans on staff. Though the early changes will appear underwhelming and clumsy, by 2025 they will seem obvious, inevitable and well beyond the changes we might envision today. Expect today’s expert doctors to think these changes are implausible when they are asked about this possibility, and expect the classic response of “human judgment will not be replaced by technology” from people who are not qualified to judge what software technology in 2030 might be capable of. Of course, the possibility also exists that a much more cooperative system leveraging both humans and technological systems in their respective strengths may also evolve, as proposed in the book Race Against the Machine, but the core functions necessary for complex diagnoses will more than likely be driven by machine judgment instead of human judgment.
The transition will happen in fits and starts along different pathways with many course corrections, steps backward and mistakes as we figure out the best approach. Given the importance of having clarity on what I hypothesize as my forecasts, I want to be clear that they are only directional guesses rather than precise predictions. Further, though many different disciplines will contribute to the innovation in medicine like biological research or new device development, I am mostly concerned with the contributions of digital health technologies to medical innovation. This should not be underemphasized, as these contributions, though potentially the most significant, are also the most variable, and hardest to predict in direction, timelines and scope.
The Fidelity of Science
Healthcare today is often the “practice of medicine” rather than the “science of medicine”. In the worst cases of the practice of medicine, medical doctors (MD’s as they are called in the US) take moderately educated shots-in-the-dark when it comes to patient care. The future of healthcare should have an approach much more akin to the scientific method, with increase data collection, analysis and experimentation to rapidly make progress. Physicians could be much more scientific and data-driven with better systems assisting them in the future, but with the increasing amount of data and research released every year, it’s hard for the average physician to keep up without technology.
The next generation of medicine will arrive at scientific and data-driven diagnostic and treatment conclusions based on more complete testing of what’s actually going on in a patient’s body.
In the past, the data to make more rigorous and scientific conclusions has simply not been available. And as a result, medical literature is rife with studies about how the practice of medicine does not meet expectations for what would constitute sufficient, correct care. There are plenty of examples that illustrate this, but they tend to share the same themes. (1) Purported experts in their respective fields frequently disagree on the effects of basic procedures. A study on colon cancer experts, for example, showed that there was full distribution across the board (0%-100%) on how valuable colon cancer screening is. (2) Things that are treated as medical fact often end up being completely wrong (yet linger for a while). Prescriptions for antipyretics such as aspirin are typically given to individuals with fever (and have been for over a century). Yet recent studies showcase that prescribing antipyretics to reduce a fever could be significantly more risky than just allowing the fever to run its course (i.e. do nothing)!
Bringing science and data to medicine will enable us to more rapidly change these debilitating problems. If we are able to collect exponentially more data, as well as collect it continuously, we will have the proper inputs to drive change. The use of data science in particular will help add meaning to all of this collected data, and over time, two very distinct improvements will happen: (i) a better validation of what we accept in medical practice about today’s therapies, prescriptions and procedures; and (ii) the invention of brand new prescriptions, therapies and procedures based on new and more holistic data about a patient.
This does not imply that the biological sciences will not be important, as fundamental scientific research in biology will keep improving our understanding of biological systems and will feed into the complex data science systems. The time period for such a scenario driven by digital technologies could be fifteen years and may take an extra decade or even more but to me, timelines seem to be far less important than the directionality. I also suspect because of reasons related to the “nature of science”, the innovation cycles for biological science-based contributions to medicine will be longer than those for the digital sciences.
Innovation Emerging out of Complexity
The healthcare transition will start incrementally and develop slowly in sophistication, much like a great MD who starts with seven years of medical school and then spends a decade training with the best practitioners by watching, learning and experiencing. Expect many laughing-stock attempts by “toddler computer systems” early in their evolution and learning; they will be the butt of jokes from many writers and doctors. Early printers, typically the dot matrix variety, the toddler generation of computer printers, did not exactly cut it for business correspondence, let alone replace traditional typewrites. But within a few generations the IBM Selectric typewriter was replaced by constantly improving printing technology. Just because a three-year-old child makes some laughable errors does not imply that they will make the same errors as a 21 or 40 year old! Similarly to equate early “toddler digital health systems” to what might eventually be possible is naive. Yet I imagine within a few iterations of these systems, we will have a world in where doctors will have much more data fed into their decision-making using the personalized medical equivalent of a Bloomberg financial terminal.
Check out this Ted talk on the subject:
It’s just that we wont let the initial toddler systems actually make real decisions while they are learning and growing up in sophistication. They will be in “assist, learn and amplify” mode with new generation of systems being developed every two or three years (a typical development cycle for a sophisticated software system) and with radical improvements in sophistication and capability over five to seven generations, much like the cellphone of 1986 (a floor mounted device for your car with heavy handset cords) grew up to be the iPhone of today! This cellphone analogy is one we will return to again multiple times to illustrate how change can happen.
The transition to the automated science of medicine will likely occur in an organic process of trial and error, starting with initial technologies and ideas that go through multiple iterations over the coming years.
Replacing 80-percent of what doctors do?
Technology makes up for human deficiencies and amplifies human strengths – doctors and even other less-trained medical professionals could do so much more than they do now. Today’s diagnostic error rate in medical practice is roughly the equivalent of Google’s driverless car having one accident per week; while this would be unacceptable for automated cars, this type of failure rate is permissible in healthcare. In fifteen years, data will transform diagnostics, to the point where automated systems may displace up to 80-percent of physicians’ standard work. Technological developments will AMPLIFY physicians’ abilities by arming them with more complete, synthesized and up-to-date research data, all of which will lead to better patient outcomes.
Computers are much better than people at organizing ,recalling, and synthesing complex information.
This will result in far fewer mistakes and biases than a hot shot MD from Harvard, let alone the average (or median for those statistically inclined) doctor I am concerned with here.
I am not suggesting that every physician will change how they practice medicine in fifteen or so years, but rather that the thought leaders will be doing so, and the future direction of medicine will be self- evident and the advantages to patient outcomes will be mostly established in well documented studies. We are already seeing this shift happening, with people on the fringe of medicine (such as myself) as well as a few (but growing number of) thought leaders entrenched in medicine taking steps to enable this future. They will gravitate to a world where the best strengths of humans and doctors are harnessed in taking care of patients, while “Dr. Algorithm” systems will work with them to do the bulk of what we know of as diagnostic and prescription work, improving via both automatic feedback mechanisms as well as human scientific input. These medical and technology leaders will show markedly better care and treatment results, and over time, the rest of the world will join the science of medicine.
Sources, timing, incentives, and pitfalls of healthcare innovation
The major problems in healthcare are systemic, and do not have to do with doctors, many of whom are accomplished, caring, honest and compassionate providers. The first problem is that globally, there is a misalignment in incentives, where organizations try to maximize revenue (extra surgeries anyone?) at the expense of optimizing care (just like some car mechanics!). These lead to hidden biases in how we administer care, and has been particularly showcased in the US in its ongoing struggle for large-scale governmental health care reform. The second problem deals with the crawling pace of change in how the AVERAGE doctor operates and gathers new knowledge, even in the presence of a rapid increase of data and knowledge about how to improve care and treatment. There is an incredible increase in the amount and complexity of newly enabled data, vast amounts of research, longitudinal health records, and medical histories. On top of this, new sensors and testing will allow for much more integrative analysis than is currently possible (especially by humans).
Utilizing this data will enable much better and more holistic care that will only get progressively better with time, yet are barely prioritized with current healthcare incentives. These lead to my belief that innovation will most likely come from outside the system.
And it’s actually relatively standard for deep innovation to happen outside of their traditional ecosystems. In most areas this happens from innovators outside the system, acting somewhat naively, failing and then realizing they need some knowledge and collaboration with the system. Entrepreneurial teams often add domain expertise to their naïve “fresh piece of paper” re-invention ideas. Society generally tries to assign more power to larger entities, like governmental institutions and the Fortune 500 behemoths, but true radical innovation seldom comes from them. Did Walmart reinvent retail or Amazon? Did General Motors reinvent electric cars or Tesla? Did SpaceX reinvent space launches or NASA and Lockheed Martin? Most importantly did big pharmaceutical companies reinvent biotechnology pharmaceuticals or did Genentech?
If it’s outside the system – could innovation come top-down from governments? Typically, growth and innovation tends to be organic for systems that are data-driven and consumer-driven. And typical lifecycles of innovation for digital technologies are much shorter than those of widespread regulation. So even if the U.S. Food and Drug Administration (FDA) or the U.S. government can help by being progressive and helping align incentives, technology (even in it’s early iterations) will be able to innovate at a faster pace. In fact, one of the biggest risks in slowing medical innovation is slowness or damaging policies by government agencies. In the worst case, this will bring some forms of technology-driven medical innovation to a halt, but it’s also likely that the innovation would just move to more progressive countries that allow for greater experimentation and use of data-driven systems. Technology that helps save costs in a first-order way, as well as technologies that have a strong mobile component can spur this non-US growth in innovation.
There are a lot of improbable sounding possibilities on how data and consumer-driven systems will transform healthcare. Though any particular one is unlikely to become a reality, it will be some improbability that will determine the future of health care as it is driven, molded and transformed by digital health technologies. Some improbable scenario today will become tomorrow’s reality. These are not absolutes but rather “more true than not” speculations. We just have to imagine what might be possible! And we must then have the courage to make those possibilities a reality.