In a nutshell, we want to help people maximize their health and performance (both how much they can achieve, and how easy it is to achieve it). Given this forum houses a lot of lifespan and healthspan extension enthusiasts, we wanted to tap into the community’s thinking to inform our approach so that whatever we build is as helpful for the users as possible.
Our idea rests on three key assumptions:
Demand exists - many people want to live longer, healthier and transcend the limitations of their biology
Supply exists - many interventions scientifically proven to improve lifespan and healthspan exist
Demand and supply do not meet properly due to significant friction created by two user pain points:
a. Decision making: choosing the right interventions is hard and requires a lot of research or a background in the subject. Either one has to spend 100s and 1000s of hours going through all the science of pay a lot of money to true experts (of which there are few and far between)
b. Execution: sticking to interventions and routines requires motivation and willpower which most humans tend to lack
To help users overcome these two pain points, we want to give them a tool that:
Solves the pain point of decision making by taking their inputs (steps, BP, heart rate, bloodwork, 23andMe data, sleep, exercise etc.) and providing them simple, concise and comprehensive personalized recommendations on lifestyle, diet, supplements, exercise and other interventions. This would be done by an algorithm using the latest science to derive implications of their inputs.
Alleviates the pain point of putting it into action through gamification (including estimating the healthspan / lifespan benefit they can and are achieving, badges for sticking to interventions etc.), reminders etc.
This tool is an iPhone app doing both #1 and #2.
As we are in early stages of development, what would be most helpful for us is:
If you could see yourself using something like this - or you think there is a better way to do this - we would love to get 15-30 min of your time for an interview to inform our approach. Interviewees will get priority for invites to our beta test!
If you have seen similar (or not so similar) solutions to the same problem, will be great if you can share the examples, whether you are using them, and what works and doesn’t work about these solutions
Thank you so much for reading!
P.S.: Apologies if I posted this in the wrong section - in which case please move to the right one.
Everyone here wants an app that takes their data and comes up with convenient, actionable insights!
The problem is that most of us have used at least a couple of apps that promised to do just that, but didn’t deliver much beyond generic advice like “get regular sleep” and “eat your veggies”.
To overcome the skepticism, it might help to be more specific, and say what you think you can do that hasn’t been tried before, and how you intend to do it. At the very least you should have something to show that helped you figure out something about yourself.
Thanks a lot for your reply Eric! Key things we are trying to do which we don’t think have been done before (at least we haven’t seen an app or indeed any tool which does this in a quick and convenient way):
Start with a comprehensive list of potential recommendations. Through a review of scientific literature we have come up with >60 interventions for which human studies (or robust animal model studies coupled with solid theoretical reasoning explaining why it is likely to apply to humans) demonstrate lifespan and/or healthspan benefits. And of course it includes things like sleep and veggies but also slightly less intuitive things such as specific diets tailored to specific genes and activity levels, specific micronutrients to get tested for given genetic predisposition to deficiencies, specific ways to eliminate specific toxins you might be more vulnerable to etc.
Personalize recommendations based on the data user has available, from activity levers, blood pressure and other things in Apple Health to sleep data, 23andMe and a quick behavioral and health questionnaire.
Prioritize interventions through rough quantification of benefit in QALY - of course it will never be precise but it is much easier to prioritize your interventions based on even a rough sizing using things like Kaplan-Meier function rather than just a laundry list of things
Does that sound useful? Why or why not? Also any examples of people trying to build tools to do something similar are very welcome.
This certainly sounds useful, though from my experience with Arivale, data-driven insights based on your genetics, bloodwork, microbiome and activity data are mildly interesting at best, at this point (hence their focus on coaching to keep people in the program, while they collect data that will eventually allow them to provide better recommendations).
Less ambitious services can be useful, too, if they focus on one specific issue (e.g. Sleepio).
Personal health data is complicated (poorly understood, and always very incomplete), so if you’re hoping for something “quick and convenient” you’re bound to be let down.
Thanks a lot Eric, this is actually a super helpful pointer to Arivale! Looks like a very interesting app. I tend to agree with you regarding “mildly interesting at best” results from genetic testing (except for some specific highly predictive things e.g. BRCA1), but bloodwork can actually tell you a lot including things at the overlap of Venn diagram of “highly predictive of important health risks” and “not routinely tested for”, e.g. Lp(a) and Apo(B) for cardiovascular disease risk.
I’m very curious to see if you can get some good use out of our forum. It’s always tricky with new toolmakers, because the participants here have a higher-than-average experience level and have seen many apps and sensors come and go, have built their own apps and sensors, been kickstarter backers of tools that disappeared, and contributors to Forum topics around keeping certain sensors alive after the companies failed. So a suggestion like yours, made within hours of joining the Forum, is naturally going to be greeted with skeptical silence. Eric’s answers are very generous, and I’m glad he piped up and encouraged me to post also. If you can stick it out, pose specific questions, and - most importantly - share what you are learning as you develop your plans, you will find extremely good feedback here.
Thank you Gary and Eric! Agreed - the collective experience of users of this forum can be an extremely valuable source of feedback and insight for building our app. Hence this thread.
As you encouraged specific questions (both for yourselves but also for all community members; would be especially interested in views of those who have used Arivale): using Arivale as a reference point:
Would you consider the outputs they provide insightful / valuable output? There are definitely useful things to test for with regular bloodwork that a standard blood panel does not (e.g. Apo(b), Lp(a) as discussed above, as well as inflammation markers, nutritional status on key micronutrients etc.), but do they do it? And do you consider this valuable or is it easier just to order these tests yourself?
Their onboarding experience is by definition not quick as you have to pick a coach, get blood drawn etc. Would it be useful to have an app which would try to offer a “quantum of utility” with whatever the user has - Apple Health data, genetic data etc. - and then encourage user to take targeted tests based on risks identified from this initial data?
The “outputs” Arivale provides are mostly just the test results; a coach then reviews the results with you to come up with recommendations. Other services like InsideTracker, WellnessFX, and Viome have automated the recommendation process, but they work with fewer dimensions, and leave figuring out how to convert the recommendations into something actionable up to you.
Arivale’s blood panel currently includes 60+ biomarkers, including inflammation markers, but not ApoB or Lp(a). I have used Life Extension in the past to do a few additional tests that were missing, or not quite frequent enough for an experiment I was doing.
Arivale’s business plan appears to depend on keeping customers for a long period of time, so they can eventually correlate genetics, biomarkers and lifestyle with outcomes. Quick onboarding and being able to provide instant insights therefore probably isn’t a priority for them. Also, data from different sources isn’t always as comparable as it should be, so they might prefer having everyone go through the same lab etc.
Thank you Eric, these are super helpful insights and pointers! I have looked into the 4 apps you have referenced and identified 5 main weaknesses / unsolved pain points that we could improve to create a better solution:
A lot of time and effort to get to first quantum of utility (subscribe, pay, go to lab to draw blood etc.)
Quality of automated recommendations low (they feel random and piecemeal rather than wholistic - especially WellnessFX; InsideTracker is better than others, seems more comprehensive on diet and something on exercise but still light on other lifestyle interventions, does not consider more advanced (and very important) inputs such Apo(B), Lp(a), toxins such as mercury
Automated recommendations scope limited (mostly diet and some exercise for InsideTracker, mostly supplements for WellnessFX)
Automated recommendations not user friendly: not prioritised, long lists, structured by input rather than result or importance, hard to choose, benefit not quantified
Automated behavioural implementation support limited (e.g. gamification, reminders helping translate recommendations into actions and put them into practice)
Do these feel like the right priorities to you? What else stands in the way of having a solution which would be truly useful for enthusiasts to optimize their health, wellness, fitness and performance?
For “enthusiasts”, practical obstacles such as the expense of specific tests (or even just being able to get their data) are probably bigger issues than the lack of instant gratification or concrete recommendations.
People who just want to be healthier (or fit into their old pants) may be better off finding a coach who can help them figure out how to make sustainable improvements. But qualified health coaches aren’t cheap, hence the interest in being able to provide automated recommendations. Brook Health is another interesting contestant in this space (they try to combine chatbots with human experts to fill in the gaps).
I’m wondering what you are looking to do for your minimally viable product and how you plan to extend it from there.
To me, automated recommendations are a secondary part of things and will need to be very specific for me to care about that “feature”. I’ve found most recommendations from others to be on the same level of meaningfulness as going to a fortune teller. “You will meet a tall dark stranger in the near future” and “you will do well on a diet of a moderate amount of carbs and fat”. If you can offer me advice along the lines of “my lower levels of copper put me in a higher risk for aneurysms (common in my family)”, the recommendations might be good.
My most critical need is a place to aggregate very, very disparate data sets (heart rate, continuous glucose monitor, sleep data, sunlight data, food I’ve eaten, supplements I’ve taken, exercise, red light sessions, sauna sessions, cryotherapy sessions, personal status assessments (pain, movement flexibility, etc), and other related items.)
Then I would like to do correlation analyses time-wise. (with these supplements and this food, my blood sugar average stayed within this target, over the next xx hours) or (I took this supplement and pain levels did not exceed xx in the next yy hours).
I need a meta-aggregation and correlation analysis app.
Given the number of different variables and limited amount of data, it’s unlikely that you’d discover a (non-obvious) correlation. But if you make some effort to keep things constant and vary a single variable (e.g. the supplement), you should at least be able to tell if there was a significant difference, or not.
Figuring out what to measure and how, collecting the data, and doing the analysis is still very much DIY, but at least the resources you have available (incl this forum ) keep improving.
This is a very interesting thread. I often found myself nodding in approval to the responses.
I like this thread for many reasons, in particular it confirms what I am seeing in other digital innovation circles - that latest gadget based data collection and analysis is over-hyped. In other words, while it is true that today’s technology enables us to go granular, in real-time on the data collection, or to “predict” phenomenon without any significant know-how in the domain of interest (i.e. thank you machine learning), we are still craving practical and action-oriented advice that stands apart from what we already know (explicitly or implicitly.)
I experienced this with the self-tracking I have been doing for the past five years. I track variables including Church attendance, time spent at the dinner table, various health, financial and academic observations and more. As an IT person, my initial approach to self-tracking was to swing the proverbial “sledge hammer” of technology in search of a “nail”. I went from spending time on the technology part, to spending more time on the experimentation part. This led me interesting experiments that, at the end of the day, confirmed what I probably already knew. Nowadays, I find myself spending less time refining the technology or data experimentation part and more time on the domain-specific research.
While the current limitations in DIY self-tracking are increasingly clear to all of us, It also feels we are on the right track and the overall trends will help to overcome these limitations. Some themes that can help in this regard are:
Focusing on capabilities instead of apps - A capability-centric view promotes the best of people, data and technology working to achieve desired outcomes.
Focusing on data interoperability - To help self-trackers share their data with domain experts; this is data assisting domain experts in providing practical advice
Thanks for the advice. Perhaps I didn’t communicate this well enough. If I had a single variable that I wanted to test to see what outputs are affected, I can understand your approach. However, I’m looking to identify which inputs generate a given output. There are far too many inputs to test each one. (20+ food groups, possible 40+ supplements, numbers of sequences of life events, etc). Moving that many items independently would take months to years of time to isolate.
I’m first looking to identify what moves the needle on a given outcome in aggregate, so that I can then narrow down the potential item from that aggregate set of items. So my goal is to track everything that I can to see if I can identify which of the many inputs moves the needle, then I can drill into refining in more detail.
If it helps, here is the underlying use case, I have an autoimmune thyroiditis condition. Theoretically, this is often caused by leaky gut. However, after reducing my food choices to next to nothing for over a month, I have not found a correlation.
I have 4 possible ways of identifying the spike of associated autoimmune response. (This isn’t a continuous thing that I can measure a small but noticeable decline and work from there - it spikes and goes back to normal.) A heart rate monitor is a moderate proxy for the hyperthyroid adrenaline spikes that seem to occur, but it isn’t a matter that I eat a food and 30 minutes or 2 hours later, the heart rate spikes. So I need something that tracks things a bit longer. I’m also trying CGM to see if the lectins in some foods may cause abnormal glucose responses as is implied by the Perfect Health Diet test protocol. This one is mostly grasping. I can also indirectly measure vagal distress (what I believe to be the most consistent symptom) and I can directly measure when I start seeing double.
Now there is the possibility that this autoimmune issue is not a leaky gut issue at all given the low level of ability to get a confirmed correlation after a month of radically limited diet. I’m also not finding a direct correlation between inflammation and autoimmune response. So my goal is to capture as much of my life as possible, along with the 4 measurements that may indicate that an autoimmune spike is in progress and then look back 2-3 days worth of data over multiple of these and see what was happening during that time period.
Then once I get something that is somewhat more repeatable, I can start isolating the inputs identified within the preceding 2-3 days worth. If this isn’t going to work, I’d appreciate any suggestions on how to not go through this with brute force sequential testing. The probability of a quick resolution has already come and gone by my following all of the suggestions of my alternative MD for an extended period of time with no success.
If there was an clear trigger for those “spikes”, you’d probably have figured out what the trigger is by now, or at least have some suspicions… So I really think the best you can do is to try one plausible intervention at a time (which could be a single supplement, diet, or a combination of multiple things) while doing all the tracking.