Supporting clinical research; a conversation with LifeQ

The age of biometric data is upon us, but the science is not ready to explain what the data means. In some ways it is really exciting, the potential is huge- even if we ignore the marketing materials and focus on the potential long term use of simple data collected under real world conditions. Stephanie Lee from buzzfeed had a sobering analysis of Apple's new Researchkit that the healthcare and clinical research value of the data is pretty much zero. I completely agree with (see my thoughts on Apple's foray into healthcare here). The only group that might see some value is the same group that has access to healthcare and quality jobs (see here for primary data from Pew Institute). This means that the biometric data pulled from "iEcosystem" will not reflect the population that acutely needs to be understood biometrically. (I'll provide a detailed example of the issues later in this blog.)

In my opinion; any data that is tied to a specific mobile device or "Internet of Things" (IoT) object is useless for healthcare unless it can be compared and combined on aggregate across devices and demographics.

It reminds of the mid-nineties when genomic sequencing was going to revolutionize healthcare and disease treatment. Twenty years later and we are finally realizing that the genome is an almost irrelevant piece. That the context of how that genome is read, acted upon by the cell, and communicated between cells is more important then any point mutations or small scale genomic changes. (I have written about thishere, in the context of cancer.)

The genomic age was necessary to spark the discoveries that are starting to change healthcare but the changes in healthcare won't be realized because of genomic biology. It seems to me that we are at the same crossroad with the Internet of Things (IoT). The technology is really cool and the visualizations are solid but......what do I do with it?

For example, I have a Fitbit it has literally change my activity due purely to trying to get to 10,000 steps......I think thats a good thing- I mean I lost weight, my back is better.....but I am left wanting more, what types of activity are related to my weight loss? Have I gone far enough-am I at lower risk for all of the things that I worry about from a health perspective?

I can tell you the data collected by my Fitbit is pretty useless to answer these questions. I downloaded it all ran it through a few different statistical models and guess what? None of it appears to be relevant to my on-going good health. I still use my Fitbit to track my activity but I have no illusions about the role that the collected data plays in my healthcare decisions.

I recently had a chance to talk to a really interesting start-up company called LifeQ. LifeQ (@LifeQinc) has restored some of my enthusiasm for IoT and real impactful changes in healthcare. LifeQ has taken a different approach to the internet of things. LifeQ owns intellectual property on for an optical sensor that uses light waves to penetrate the surface of the skin to monitor multiple biological measurables; heart rate, blood pressure, oxygen saturation, with other important measurables such as glucose in the beta testing phase. The real power of LifeQ is not the measurables. Most of the metrics that their sensor measures are relativelty common place. Many devices can measure heart rate, blood pressure, glucose, these are not unique. The true value of LifeQ as a IoT vendor is really in the predicative models and software that allows identification of changes in ones own biology. As Christopher Rimmer pointed out this very similar to the model that Google, Microsoft and Apple have pioneeered. LifeQ owns the core data acquisition ("OS") and the core platform for integrating and using the information ("search engine"). If LifeQ can be half as disruptive in healthcare as Google has been in mobile, they can be a driving force for systemic cost reductions and better treatment outcomes.

The device agnostic approach gives LifeQ a wide potential market in healthcare, fitness as well as the flexibility to weather the inevitable changes to the device ecosystem that end users are willing to use. The focus on data acquisition and analysis reduces the overhead and ensures that the width and breadth of data needed for accurate modeling can be gathered.

As LifeQ told me during our conversation "You can't build a great, high quality algorithm and data access AND build multi-functional devices at the level required to collect the data we need. There are plenty of companies in the health, medical and consumer device world with the pockets and desire to build high quality readers."

It is a really smart strategy, especially in the complex global healthcare and lifestyle market(s). Focusing on their strength and being choosy about the partnerships. This strategy allows LifeQ to ensure data quality and more importantly from a medical perspective, information security. High quality data that is combinable across devices is necessary to keep the predicative models relevant, and increase in accuracy with successive iterations.

Obviously the key risks are in how to ensure the partners continue to innovate on the physical devices and the integration of different device collected data into a single model. To keep with the Google analogy how do you build the back end to protect against the fragmentation of the device type when each device manufacturer has specific needs and market segments. The kind of companies that they are dealing with understand the necessity of spending on the hardware.

Not surprisingly the initial partnerships are consumer focused, within the personal potential niche, for example those that cater to extreme athletes. Some wider consumer focused. An interesting aspect will be how LifeQ can integrate the niche data into the predicative model without biasing against normal peoples fluctuations. For example, we know that part of what makes elite atheletes, well elite, is speed of recovery; their heartbeat decreases at rest faster, their rate of breathing decreases faster, muscles recover faster. So as LifeQ collects this data, what value does this data have for us "normals" will the models be accurate?
It is not an insurmountable challenge but the awareness of how the data can influence the model and vice versa is a concern for any IoT or Quantified Self technology. It is the early adopter problem, your initial feedback from fanboys and people who share your vision can blind to the general publics use cases and expectations. It is the exact problem that caused Google to shutdown Glass.LifeQ seems quite aware of the potential founder effect problems.

A more important (to me at least) is that they are also engaging the medical community to enable the kinds of use cases that have long term quality of life and better diagnostic test values for health monitoring. These kinds of markets are a growth market and can provide a reliable revenue stream. For example, at home monitoring or ambulatory care for basic monitoring of HR, breathing, Oxygen levels, blood glucose (coming soon). All of which can be monitored today by the LifeQ powered devices. The problem being that the current monitors that have the accuracy that LifeQ needs are cumbersome but they are easier to where than what most hospitals have- and can be worn for long periods of without patients being strapped to wires or stuck in bed. The potential for clearer test results under real conditions is tantalizing.

What is next?
Like all start-ups LifeQ is focusing on ensuring their product is the best by ensuring that every element that could negatively affect its core product. The really interesting piece will come from the meta-analysis once the number of users hits a large enough N to ensure predictability across populations.

LifeQ acknowledged the potential limitations of a optical sensor; skin color, lean muscle to fat ratios, as well as stability issues cause be user activity. They are working expanding the repetoire of sensors that LifeQ can collect data from as part of the platform.

The real issue that faces LifeQ and any of the more robust quantified self devices and analysis platforms really comes down to action steps. For that matter the same issue exists for personal genomics. What is the line between variation of the population and dangerous biometric signature? Is there more harm then good from telling folks everything?

LifeQ has a great platform, and appears to have all the pieces in place to be the "Google for healthcare." They certainly bear keeping an eye to see what they do next