I have blogged a number of times about predictive analytics (see, for example: Identifying Patients for Remote Monitoring with Predictive Analytics; Eric Schmidt Discusses the Potential Value of Predictive Analytics in the ER). Much of the hype about this technology and set of products is largely theoretical at this time but great potential looms in the future. I therefore was very interested in a recent article that provided a very pragmatic perspective on the topic (see: Big data and predictive analytics pull in smokers for lung screening). Below is an excerpt from it:
Virginia-based Chesapeake Regional Healthcare sought to motivate smokers and former smokers to "Get Off their Butts" and get screened with its Lung Cancer Screening Trigger Campaign, which they started in 2015. By leveraging SaaS-based big data analytics and marketing technology, Chesapeake Regional Healthcare was able to identify, target and educate certain populations at risk or eligible for lung screenings with this personalized outreach program. As a result, the healthcare organization was able to get 5.21 percent of new patients and 9.17 percent of all patients it targeted to get lung screenings. "Chesapeake Regional Healthcare was looking to drive more volume into our Lung Cancer Screening program, and only certain people need or qualify for a lung screening – targeting and predictive analytics are perfect for this type of campaign," said Sarah Liebrum, manager of marketing and communications at Chesapeake Regional Healthcare. So the health system turned to Tea Leaves Health, a developer of cloud-based analytics technology that Liebrum said "offers self-reported smoker data as well as some modeling that allows us to predict who is likely to need this screening."
To get right to the point, a healthcare system "was able to identify, target and educate certain populations at risk or eligible for lung screening with...[a] personalized outreach program." A very high percentages of patients it contacted scheduled lung screenings with a large number of cases of lung cancer identified. Campaigns like this take the theoretical advantages of predictive analytics to a practical level. I suspect that many patients who were long-time smokers were not inclined to ignore a letter from their hospital indicating that a "computer study" indicated that they were at a high risk of lung cancer and should schedule a diagnostic screening. Even if the male recipients of such a letter were still inclined to ignore the advice, the letter was probably opened and read by their wives.
I believe that one of the reasons why this example of predictive analytics was so successful was that the participating health system, Chesapeake Regional Healthcare: (a) could pursue an admirable public health objective, screen for lung cancer; and (b) simultaneously generate revenue-producing activities in terms of diagnostic procures and, for some patients, surgery and disease treatment. In short, health screening is more apt to be supported by health systems if it produces subsequent business and revenue for them.