A short essay written back in Fall 2018, ideating around potential product & technology approaches to shifting healthcare to a preventative system, both for micro and macro level care. The post itself is a little naive and underdeveloped, but it is interesting to see that many companies have popped up to tackle these exact problems, and we've also gone through a Pandemic level event (COVID-19), something this essay included in its consideration as well.
It’s fairly well known that Modern Medicine has been built on a crumbling foundation: the periodic checkup. Health evaluations like this don’t work well, largely because most deadly diseases don’t really rear their heads until the patient is symptomatic. This is also when they are least treatable. Inefficiency is inherent to the checkup because the procedure and practice has its roots in early medicine, where physicians couldn’t effectively diagnose because of a lack of noninvasive clarity and less knowledge about what they were dealing with.
It is 2018 now, and many continue to scratch their head. Why haven’t treatable diseases been largely eradicated as a contributor to overall mortality? There is a clear lack of incentives in the healthcare systems that exist, despite our clearly advancing capability. Thousands of variables track things like the advertisements we are likely to click on, why do we settle when it comes to our health? These problems only persist when you look at it beyond the individual scale, with large populations of underserved people frequently subjected to enormous public health problems like epidemics on a startlingly regular basis because of spotty data reporting and lack of application of technology used in other spaces.
Human biology is really a hierarchical system that is both dynamic and emergent, and the known unknowns are ever increasing, yet we’re content to take a few single variable snapshots over years. It seems ludicrous.
It should be possible to build a multilayered, preventative healthcare system that fits the individual and continues to cheapen and gain in efficiency over time as a result of data network effects. I am certainly no visionary in proposing this, nor am I claiming to have a solution to these fundamental problems. In fact, this sort of exposition has been presented by many brilliant folks, from healthcare professionals and academics to entrepreneurs. Rather, all I’m introducing is a simple framework to consider the various levels of effective preventative health and observe how they operate in a systems stack and inform each other to hopefully increase the baseline of quality of life and health for every single one of us.
Following an explanation of the overall framework, I briefly outline the specific areas on which I will be specifically focusing myself over the next few years, in what should hopefully be an effective execution to achieve what I describe.
A master plan, if you will.
A Multi-tiered Framework for Preventative Health
The fundamental focus of this framework doesn’t center on the logistics of data ownership and implementing interface between all the fragmented providers, tools, data types, and patient groups. While this is certainly a difficult problem, it seems to be something that many brilliant folks are focusing on, and while it is a fundamental bottleneck to much of the proposed activity that will follow, I think many can and will solve this problem within the next few years. It may even be an incumbent, as it seems to be a largely logistical issue. I don’t generally reason by analogy, but it can probably be likened to a big data vs data insight issue. I’m personally much more inspired by what we will do with the data that becomes standardized, though some improvements in directed data collection may be necessary to effectively drive the first few steps over the next few years.
1: Macro-trend prediction and public health forecasting
Global disease burden is overwhelmingly costly to humanity. The direct cost of Malaria treatment alone is more than $15 billion/year. With forecasting based prevention we can save millions of lives, billions of dollars, and improve human productivity
Building early warning systems and forecasting the impact and spread of public health problems like disease is not a novel concept and some have already been implemented to great success. However, the stream of data that is now available, including the addition of high quality Space-Earth observation data provides an excellent foundation for the introduction of more nuanced and advanced ML based solutions. Accuracy in the ballpark of 90% has already been claimed when it comes to forecasting specific diseases like dengue fever by some groups, and the use of deep neural nets has also proved a fruitful endeavor.
The real moonshot would be generalizing the technology and applying it to a plethora of public health problems, including disease, famine, and environmental dangers. The audience that would likely greatly benefit from this are underserved populations where these sorts of problems claim upwards of millions of lives a year.
Based on the research being done in this space, significant value can probably already be delivered to various organization types, from governments to individual hospital systems.
2: Individual Preventative Care
Perhaps the most popular and familiar, building personal preventative health systems that monitor your well being after consolidating your clinical health data — i.e. Electronic Medical Records, Genetic and Lab Reports, and your daily data — i.e. fitness, sleep, and diet, would be enormously helpful in building a longitudinal observation system and improving our overall health. There are several teams focusing on this, or some problem upstream to this, though progress still seems to be minimal. It should of course be accompanied with the ability to own your data, rather than some provider
True integration and prevention would likely manifest itself in preventative suggestions rather than diagnosis, though apps like ADA health have made great progress in the latter. Think notifications telling you everything from where more sunscreen today to make sure to eat some food high in omega 3s this week, like salmon.
The possibilities and implications are wild, and together with the Macro level prediction, these two improvements alone promise significant upside with regards to health and productivity gains at a population or civilization wide level.
3: Inpatient diagnosis and treatment
There is a lot that has already been done in this space, but much that can be improved with the first two tiers of the framework in place. Inpatient treatment will largely mimic stage 2, while diagnosis will directly improve as a result of the data network effects in place with both prior points.
Alternative preventative therapies are an interesting area to explore, and there are various groups working on leveraging expressive arts, virtual reality, and other technologies to great effect.
4: Bioengineering of time-sensitive therapeutic genetic circuits.
Enormous advancements in bioengineering are taking place, with some teams building computational environments that allow you to design and select for an optimal specific genetic circuits. A professor of mine, Dr. Peter Girguis at Harvard University, once explained to me how powerful biologically integrated therapeutics could be for preventative health, where timed or sensor specific genetic circuits release some specific molecule that treats a particular condition or malformed pathway.
Step 4 is advancing fast, but will really be supercharged by the plethora of longitudinal data that is provided in stage 1 and 2, similar to the effects that drive stage 3.
Together, I think these levels constitute a systems level implementation of preventative health that should hopefully drive progress in human health and shift care from treatment to prevention, saving millions of lives and billions of dollars. The end result of this system would be our knowledge approaching large scale quantification of genetic and physiological dynamics in the human body, and is something I’d like to see happen by the end of the century.
I have no doubt that there are many brilliant minds working in this space, both from the top down and from the bottom up, and the real challenge I expect will be unifying fragmented, vertical specific implementations into a broader, system wide result. Incentivizing data sharing, security, and navigating other key obstacles is going to be difficult, and I’d love to hear your thoughts.
Here, I’ll outline how I’m going to contribute to this overall framework in what I hope seems to be a rational and organized method. I’m a fresh faced new grad student, so I’m sure what follows seems logistically naive, but bear with me. The plan can be surmised as follows:
1.) Work on Macro Public Health prediction — while this may seem difficult, a lot of research has been executed in this space, and it seems like a promising way to both move the baseline of human health forward worldwide while offering a concrete value proposition to organizations like governments, health insurance companies, hospital systems, NGOs, and others. I’m personally doing research in this space, leveraging Epidemiology, Machine Learning, Algebraic Topology, and Space-Earth Observation.
In packaging public health prediction and offering it for a reasonable price, I hope to generate a baseline revenue that will support the implementation of stage 2.
Targeted Customers: NGOs, Governments, Hospital Systems.
2.) Implement Individual Preventative Care — Using the work and funds generated in Step 1, I hope to implement stage 2 by building an application that allows anyone with a smartphone to receive personalized predictive health suggestions that can guide their activity on a day to day basis based on the data provided. An obstacle here is of course data consolidation and regulation around EMRs, but I believe it should be possible to drive adoption and provide value by operating on fitness, diet, sleep, and genetic data until the consolidation problem is solved.
Now, it seems enormously difficult to drive adoption here immediately, so step 1 should theoretically aid in the establishing of reputation, funding, and general machine learning development that will allow for the effective implementation of step 2.
To effectively test the use of this proposed platform in a low cost, low overhead way, I will likely begin building simple models around my own data, and build some sort of dashboard to help me quantify my own health and suggest activities that will provide preventative benefits.