One of the more profound business lessons we've learned over the past year was through talking to many different kinds of companies that rely upon disease surveillance data from the Centers for Disease Control and Prevention (CDC) and other organizations that acquire and resell laboratory data; that surprisingly, those clinical data don't correlate very well to sales and claims data for our enterprise customers. For example, there's a discrepancy between the volume of people who are diagnosed with flu in a region compared to those that actually purchase flu remedies. The hypothesis is that there are a growing number of people who remain undiagnosed, who never even go to the doctor, but will self-diagnose and purchase products based on the illness they believe they have. This set us on a path to better understand our own correlation to these sales data. The results of that analysis are published in the following case study here.
In short, our data showed a 0.9r Spearman correlation coefficient to adult allergy remedy sales and 0.8r to flu remedy sales (out of a perfect 1.0r). Given the accuracy of our flu forecasting models (and soon to be released allergy forecasting models) we now know that we can also forecast sales of allergy and flu remedies. Interestingly, our data (also supported by academic research) shows a strong correlation to the clinical data as well, and perhaps this is because we are collecting reports of those who have been to the doctor along with those described above who self-diagnose, therefore allowing us to have a more complete picture of health in any given population. Somewhere in between pure clinical data and pure sales data exists Sickweather's data, which provides a bridge of insight between the clinical and the anecdotal.
submitted by: Graham Dodge, CEO