Predictive Model Generation for Asthma Patients | STAND 8 | Stand8
We partnered with a leading healthcare provider to tackle something that affects millions of people — asthma and allergies. (While we reference both in this article somewhat interchangeably, the client in question was looking for a way to limit severe asthma and allergy attacks that are especially dangerous.)
It's estimated that anywhere from 20% - 40% of the US population suffers from asthma and allergies. According to AAFA.org, that's more than 50 million people — with allergies being the sixth leading cause of chronic illness in the U.S.
In fact, the proportion of people with asthma and allergies is increasing. (One of the leading theories on this is the "hygiene hypothesis.") If you have seasonal allergies, the only solution is to take medication. In order to be effective pre-medication, taking medication before the onset of symptoms, is necessary.
This approach is harder than it sounds. No one knows when pollen season will begin, and each person is affected by different allergens at different times.
The client needed a personalized solution for each patient seeking care. We turned to the data for answers.
The first thing our Data Services team did was seek relevant data for the 2 biggest factors in the allergy equation — patients' individual allergies AND air quality data. We combined user input from the client's healthcare management app with geospatial and temporal air quality data.
With this data we built an automated predictive model that indicates when each patient is likely to experience asthma or allergy attacks. We shared individual predictions with users so they could know when to participate in outdoor activities or when they should bring their inhalers for more acute attacks.
To further improve the predictive model, we shared user-entered symptom data with healthcare providers.
Given the scale and ubiquitous yet somehow nebulous nature of this problem, the client and their patients were exceptionally pleased with the results.
Of the 40k users of the mobile, we were able to build customized predictive models for each person seeking care for allergies and asthma.
In the first year, the testing group avoided 25 trips to the emergency room using STAND 8's predictive model which allowed them to pre-medicate or avoid higher-risk scenarios.