doraby

http://dorogenskaya.com/blog/bio-en.html

Why visualizations in Health don’t work

Visualizations in the most favorite health apps don’t have enough comparing and exploring possibilities.
The main 3 problems:
  1. Meaningless numbers.
  2. Excessive unreliability.
  3. Poor comparing possibility.
Solution — single dashboard with the custom widgets.

Meaningless numbers. 42.

Healthy lifestyle evolves from good practice into hygiene requirement. People want to get rid of the current health problems, prevent problems in the future, maintain weight balance, have a beautiful body and sleep well. The most common questions related to these main reasons.
Many apps provide you tracking tools and collect under the hood large amount of data which unfortunately does not turn into something valuable. Majority of these apps provide data visualization that should help to see the answers. Does this visualization help? Let see.
It’s the first screen of my Health application, and it should answer some questions bother me. Is my health getting better or worse?
Health pretends to be the supercomputer from the Douglas Adams Universe that answers all questions with weird numbers.
— Is my health getting better?
— 430 kcal, 81 bpm, 9 hours. 4459 steps.
What the heck do all numbers mean? Is 263 kcal good or bad, more or less? If my goal is 1000? If 2000? If it’s 9:00 in the morning? What does 263 mean? What do 4459 steps mean energy-wise? Is 7:49 enough to sleep?
I am implied to know all my numbers by heart and instantly compare them on the spot. I don’t. Health should notify me if my results are good or not sufficient. But until it is unknown by comparing with what nobody can answer this question.
1. Compare with my own latest result. I have to see my previous results at a glance.
Let say that now I’m fine and my goal to prevent problems in the future. I need to maintain my current form, so I need to know if my daily indicators match the norm.
2. Compare with my future goal.
I have to see this goal. And I couldn’t set goals without some previous results. Usually, Health applications offer to set these goal numbers from scratch. This unnatural suggestion confuses because I have not known my current performance yet. The application should track my habits automatically and suggest some goal based on my real data. Maybe that 10000 steps are not enough to me. Or vice versa, my physical form doesn’t allow me to reach these numbers in any way.
If I want to sleep better, I would prefer to compare with my goal instead of the normal state, because this “better” should be determined.
But it can work better if I save users from comparing numbers and show relations visually.
Let say, I want to have a good shape. Maybe for someone weight is an essential metric, but the low weight does not guarantee nice shape for skinny people.
I judge by circumference proportion of my Breast/Waist/Hips when I need to shape up and gain some muscles while Health provides me only the waist. So I just collect the data and visualize how it might look like.
3. Compare with the result of somebody from my friends, family, kinfolks. 
Some people like to compete with the others. It drives them and inspires to new achievement.
Personally, I hate this. I don’t like to compare even with myself. My goal is not an achievement, but the quality flow of my life. Notifications “It’s better than 75% of unknown and unimportant people” make me skeptical smirk.
The solution is simple — I should decide what exactly I want to compare with and choose key metrics. Without this simple obvious thing visualizations will always try to answer the Ultimate Question of Life, the Universe, and Everything.

Sleep trackers obsession. Excess of unreliable information.

Science researches state that all of these tracking tools are not able to reveal sleeping cycles precisely. Human sleeping behavior much more flexible than numbers on our phone. If we lack a deep phase one night, the next night it will increase.
Anyway, people suffer when sleep badly, install apps and try to guess what they have to do with a particular sleep problem, and what they are doing wrong.

One night from Pillow

How good is my sleep today?
Some numerical indicators can partially answer my question, but not precisely enough. It says I slept for 8:39. And I spent more than 9 hours in bed. Is it good or bad? I would like to see the dynamic at least during last week, but ideally for a longer period, a month or two.
Visualization of sleep session is quite good — periods are displayed as a rectangle with durations encoded by size. But as long as it tracks incorrectly, this is excess of unreliable information.
Scientific hypnogram appears more natural because it looks continuous. It shows brief awakening moments that are often too short to be noticeable.
Sleep apps only show a long perceptible awakening period and track them incorrectly.

The same night in Sleep Cycle App

Comparison capabilities in the Sleep Cycle work better. It shows weekly sleep progress.
But one week is not enough to judge the dynamics.
Is 8:49 good or bad? It says it’s “95% quality”. But Pillow says it’s only 80%. Someone lies.
One thing seriously bothers me. How is 95% calculated? Maybe it related to the percentage of deep sleep and total sleep time? I can’t understand, so the real answer to the question “How good is my sleep today?” sounds like:
— Perhaps 95 % of the unknown 100%.
Do you really need that unreliable accuracy in the answer?
I want to confess, I have never scored 100% yet. Why, if I feel pretty rested, fresh and calm it is not taken into account. I would prefer to choose my own 100% depending on my feelings. It’s bugging me and recall kinda manipulative mechanics that pushes me to proceed using this application.
I read frightening books that promised me that I would die of cancer if I didn’t sleep 8 hours today. In the end, I stopped to follow this burden of knowledge because it gives me nothing except anxiety growth. So, if you recognize yourself this visualization is for you.
We are not alone, researchers have already noticed that people who have started to track their sleep complain of its deterioration.
…So, people should use his app to look at trends, not to obsess over the daily data…
…One big problem is that many of these tracking apps send daily notifications with users’ sleep statistics. He suggested that people experimenting with sleep trackers look at weekly trends, not daily summaries.
…It’s a vicious cycle because if you immediately feel you didn’t sleep enough, it will impact your mood”
…Cleaning up your sleep pattern may be a more effective way to resolve your sleep problems than using a sleep tracker. This can be achieved by monitoring the time you get up in the morning and making sure it is regular.
I totally can relate to this. People need long term dynamics.

What visualization can work better?

I see trends. I see how new habits work. I understand how my sleep correlates with the goal. Poor sleeping stands out. BTW, it pretty often correlates with alcohol. 🍷
I left ratios of sleeping cycles and it’s stacked bars visualization deserves the accuracy it gives. Now it emphasizes noticeable outliers that correspond more to reality and have meaning only in the large massive of the data.
Area chart makes deviations more obvious, but less precise by days.
If we had a candle chart we might know even more about the core sleeping habit — sleep and wake time.

Habits and Symptoms. Sense of achievements, controlling anxiety.

Why do people track their stuff? It’s a kind of diary and mental practice to get awareness about achievements and mistakes scattered in time. Some people call this visualizations life mapping, because the behavior pattern defines the quality of life, sense of fullness and satisfaction.
Am I good today and what should I do to be better?
I’ve tried to track this stuff in different apps, and Flo seems to be the most comfortable way to do this. There is an existing report section where you can explore dependencies of symptoms and habits on the cycle. If these symptoms correlate with the cycle, it works.
How it visualized in Flo
But I would like to see mood and symptoms on a single screen, especially if they correlate to each other. And, surely, as in the sleep tracking case, one cycle is not enough to reveal trends.
I would like to know the frequency of symptoms for a month or even a year? Moreover, I would like to know the dynamics of these symptoms. Do they appear more often or less?
Actually, I would prefer to see all these events in one place on the single dashboard with the possibility to add, sort and remove different widgets.
Users will be able to assemble their dashboards from the apps they use. This all-in-one life map can aggregate all data in one place and replace separate poor charts.

Tags

Topics of interest