24Jan24 | Data Science + Health

Some thoughts I wrote down about “algorithms for self-tracking”:

1. What can’t be measured can’t be improved. So health doesn’t start with health; health starts with tracking.
2. The depth of self-tracking could be infinite. Meaning there’s a large array of things we can measure about ourselves, but even then there are multiple levels to each one of them.
3. Almost all self-tracked data is connected to each other.
4. On average, the human body generates an order of 100s of thousands to a few millions of data points per day, from which a few 10s of thousands are probably recorded on some form of a digital device, and from which even a smaller % that we can use or look at.
5. So on average, we can only make use of roughly about 0.01% of the data our bodies generate on daily basis for our betterment and wellness.
6. The first step to make sense of any data is to ask questions. Data on its own is not going to provide much value if we don’t ask the right questions. Asking these questions is what helps us get better, do better, & be better.
7. Questions can get so deep & complex that at some point along the way not all of them can easily be answered.
8. Technology represents the current state of what’s possible in terms of data storage, processing, speed, & convenience. As time passes by, technology gets better at answering more complex questions.
9. Not only that, but also as users become more aware about the importance & utility of their own data, their questions change over time with respect to their priorities & goals.
10. So not only there always exists a set of more complex questions that current tech can’t answer now but will probably do in the future using the same data, but also there always exists a set of simple questions that the user happens not to be interested about at the moment, but could probably be interested about at any point in the future.
11. What does that tell us? That capturing high-resolution data is literally the safest bet! Optimizing algorithms for a specific set of questions can get obsolete very soon, but optimizing for capturing clean, structured data is the best way to be future-proof.
12. Being future-proof means being able to answer questions of now & questions of the future. But unfortunately there’s no easy way to do that.
13. Very recently, AI assistants have been able to do all sort of tasks, even analyze & visualize huge amounts of data. In the context of self-tracking, this means that whatever data we have about ourselves can now be easily used to extract insights & get informative visualizations using a simple interface as easy as a chatbot, without a need for a human data analyst to crunch the numbers as we used to have.
14. With this combo (hi-res data + powerful AI), the possibilities are really limitless. We can easily solve the problem we discussed earlier. So it doesn’t matter if a user changed interests, goals, or priorities, their data is always going to hold the answers anyway, & interacting with this data is now becoming easier than ever with AI.
15. Whatever breakthrough comes next, it would be embedded deep in the “data processing” side of things.