Interested in learning what's next for the gaming industry? Join gaming executives to discuss emerging parts of the industry this October at GamesBeat Summit Next. Register today.
IBM today announced a new tool that taps AI to predict when allergy symptoms are likely to flare up. It’s called Allergy Insights with Watson, and it’s available in The Weather Channel app for iOS and Android ahead of a launch on the web.
In addition to a 15-day forecast that predicts allergy symptom risk (e.g., high, moderate, low) and a 3-day outlook for allergens, Allergy Insights delivers notifications when allergy risk is changing and explanations about how weather conditions can trigger symptoms. It also provides pollen levels by allergen (with mold coming soon), tips for managing allergies or reducing exposure, and news articles and editorial content related to allergies.
According to a recent survey conducted by IBM, most allergy sufferers — 60% — use weather forecasts to help manage and mitigate the worst of their symptoms. But pollen metrics like tree, grass, and ragweed levels, which the bulk of apps use to assess risk, aren’t necessarily good predictors, and their sources tend to be spotty.
That’s why IBM scientists trained the model underpinning Allergy Insights on data from IBM MarketScan, a family of anonymized health corpora representing over 100 million patients; location information; and weather attributes like temperature, humidity, rain, wind, and dew point. The geographical data enabled the model to understand what flora is growing nearby and when allergens will be produced, while removing references to the time of year helped reflect changes to the start of allergy season attributable to climate change.
The result? IBM claims Allergy Insights — whose predictions don’t reflect air quality levels — is between 20% to 50% more accurate than algorithms that take into account pollen alone. Moreover, it can predict allergy risk down to the ZIP code.
“After extensive research, pollen data and air quality levels were excluded from the predictive model, since they proved unreliable indicators of allergy risk,” said IBM. “While no two allergy sufferers are the same, knowing in advance when symptom risk might change can help anyone plan ahead and take action before symptoms may flare up … The team will continue to review pollen data and include it when it’s more reliable.”
Interestingly, it’s not the first instance of AI being applied to the problem of allergy and pollen prediction. In 2018, Doc.ai, which offers an app that connects health companies and medical researchers with smartphone users, built a model to anticipate allergy risk drawing on user data like BMI and physical activity. Separately, researchers at the University of Texas at Austin designed a device that measures pollen levels from specific locations throughout the day.
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Learn more about membership.