Decoding chronic illness
Using data from 1,700 patients to paint a picture of life with illness
Chronic autoimmune and invisible illnesses are an incredibly tough problem, involving many frustrating non-specific symptoms that may be triggered by a vast array of potential environmental and biological causes: food, stress, medications, side effects, sleep, weather, and a million other factors. No one — patient or doctor — seems to know where to start.
What if there was a tool that could record your unique symptoms and environment and help you manage them? And what if we could catalog data from thousands of patients like you to find out what triggers chronic illness and how to treat it?
I was lucky enough to find a team of friends from around the world who wanted to build that tool. It’s called Flaredown, you can start using it right now, and we’re finally getting an early glimpse of the insights it can provide.
It starts with a useful tool
Our data comes from our free and open source Flaredown app, which makes it really easy to track symptoms, treatments, and environmental factors. Over time the app will visualize your symptoms and treatments so you can understand how they interact.
Everyone’s different, so Flaredown is built to handle your unique illness. You can add any symptom if we don’t already have it, and there are no restrictions on treatments or anything else.
An early look at the data
After disregarding accounts with inadequate data, we’re looking at about 1,700 Flaredown users reporting 450,000 datapoints over the past year beginning with our private beta. Data scientists Peer Karmaus and Graydyn Young took a look at our anonymous, aggregate data and gave us an idea of what we’re all tracking.
Most of us are females from developed countries
We span 57 countries, with notably high user numbers in the US, UK, Canada and Australia. Across all regions, women are the primary users of Flaredown. That’s not surprising considering they experience some of the most common conditions in our database at much higher rates than men (unfortunately I’ve found less data on other genders). For example, women are 2–10 times more likely to experience autoimmune disease than men.
We each track a lot of conditions
Initially we expected to see people focus on a single condition and a small set of symptoms related to it, so we were surprised to see just how many different issues you’re all tracking.
On average, users indicate eight conditions at any given time. In fact it seems that chronic illness is rarely described fully by a single diagnosis, as few of our users are tracking only one. In total we’ve recorded almost nine hundred unique conditions.
In the network above, a line between two conditions shows that at least 25 users have tracked them together at some point, with thicker lines indicating more users have tracked both concurrently.
Depression and Anxiety are some of the most-tracked on Flaredown and this chart shows you why: they occur alongside so many other conditions. It’s hardly surprising that the mental burden of living with chronic illness would lead to mental distress, but it’s possible that in some cases they also stem from the same problems causing the underlying illness.
In the case of conditions like Fibromyalgia and Chronic Fatigue Syndrome it’s hard to say why they have crossover with so many illnesses — it’s possibly a result of the non-specific nature of their symptoms, which might cause them to be a match for more people than other more defined conditions.
We share many of the same symptoms
You might expect that different conditions would have a recognizably different set of symptoms, but that’s not often the case. Many symptoms occur in more than half of all conditions on Flaredown. That means no matter what illness our users track, they’re more likely than not to have experienced most of the following at some point.
This won’t surprise the many patients who’ve had to struggle to find a diagnosis when they matched with so many conditions, or who have noticed that every miracle remedy on the internet seems to fit their symptoms.
Wouldn’t it be amazing if we could figure out how to fix Fatigue? We could improve so many people’s lives all at once. Unfortunately because they can be caused by many different factors for many different people, it’s up to the individual patient to find and address the cause of these non-specific issues, which typically takes years of trial and error. We hope Flaredown will be able to change that.
We’re taking more than just prescription medications
Doctors mostly prescribe medications, but people are clearly experimenting with many other treatment options — including supplements, environmental and behavioral changes and alternative therapies.
There’s only one prescription medication on this list. That makes sense as medications tend to be more targeted — and indeed if you look at the data for a specific condition you see many more drugs in the top charts — but clearly users are experimenting with alternatives.
And why not? If you look at the treatments on the left you’ll see that they’re inexpensive, available, and have little to no side effects — some of them are good for anyone, like sleeping and exercising. It stands to reason that every patient would give them a shot at some point.
Our userbase is still small enough that a few people heavily tracking a particular treatment can skew it, so we expect this chart to shift and normalize as we grow, but it seems likely that the general trend toward non-prescription treatment popularity will continue.
Side note: three of you are tracking Zombies, Run! as a treatment. Took me a while to figure that one out, bravo.
Life with chronic illness
What do we deal with every day? Let’s take a look. This word cloud represents some of the most prominent symptoms and environmental factors that users are logging, with size representing the total amount of times tracked. You might want to zoom in if you’re on a phone.
At this high-level view we’re seeing the symptoms and environmental cues that occur across the widest variety of conditions, because no single illness is big enough to push its unique symptoms into this chart.
I think it’s interesting that we see items like “went to work”, “had sex”, “productive”, “cleaning”, and other normal life events. We don’t typically take these things into account in treatment, but users chose to log them in our symptom-tracking app because they believe they have some interaction with their illness — enough to make them some of the top items in our dataset. We should recognize the degree to which our illness is intertwined with every part of our lives.
The future of medicine
We think the medical industry is behind on adopting modern tools that have revolutionized so many other fields. Our vision is a future in which powerful algorithms use datasets like Flaredown’s to help you control your health, and help doctors understand what causes illness. In fact, we’ve already started exploring code that we hope will help us…
Understand our environment’s impact on our illness
If we know how our environment and symptoms interact, we have a better chance of controlling them. By correlating factors like sleep, stress, food, side effects, allergens and anything else that might trigger symptoms, we can find and eliminate the source of our problems.
Help doctors diagnose more accurately and quickly
We want to give doctors new tools by training an algorithm to look at all available data and calculate the likelihood of different diagnoses. Our first attempt can guess our current users’ conditions correctly more than 80% of the time (though we expect to develop it more to handle a more general population).
Recommend effective, personalized treatments
Not every patient responds to the same treatments. We think that by finding users who are similar to you and understanding what has worked for them, we might be able to recommend better treatments. We don’t want to endure years of trial and error to find an effective treatment, and we want to include all the options, not just prescription drugs.
Improve our quality of life
We visit doctors because we feel bad, not because we want to have better numbers on tests. By creating truly patient centered outcomes measures defined by the impact patients report on their well-being, we can evaluate treatment plans according to how much they improve our lives.
Want to help out?
The easiest way to help is to sign up for an account at flaredown.com and start tracking your symptoms and contributing data. Flaredown will always be free, but you can donate to the project here. And please share this article and our app with everyone who it might help.
If you are a developer (particularly for iOS), doctor, researcher, data wrangler, blogger, or otherwise want to contribute to Flaredown or help us get the word out, please send an email to contact@flaredown.com.
Illustrations by the absurdly talented Nataly Ouvarova.
