02/15/2018 in Blog
The idea of an off-the-shelf, analytic solution to a complex problem is an attractive dream that many people across a variety of industries are constantly searching for. Data science methods often present themselves as a black box, or magical tools with a limitless ability to solve the most complex problems. The idea of a "Watson" or "Siri" wading through terabytes of data and pumping out analyses, insights, and prescribed courses of action is exciting. In reality, these fantastical ideas are still visions rather than concrete tools to improve business outcomes.
So, is optimism about data science misplaced? Not at all. While reality is less flashy than an AI reminiscent of science fiction, data science does deliver value to companies across the globe. The key is finding a team that can perform the hardest task in data science: connecting analytic solutions to business needs. Many smart people can build algorithms and run statistical analyses, however a good data scientist knows how to use those tools to solve real-world problems.
For example, our team focuses on identifying the problems most important to healthcare executives, figuring out how to dimensionalize and analyze those problems, and finally applying advanced analytic methods to deliver insights and value. We call this algorithmic intelligence: human judgement, enhanced and supported by technology.
How do we apply algorithmic intelligence to a problem? Consider this example. One key question in the healthcare industry is, "How can we effectively engage consumers to meet their healthcare needs?" Let’s break this down into four key areas and apply data science approaches to effectively address each one.
Opportunity Identification: How can you identify the service lines in your facilities that are both responsive to new initiatives and highly valuable? Quantify value and responsiveness, and use data from a wide variety of sources to evaluate each service lines potential impact to your total business. This data grants hospital marketing teams the ability to target consumers that are good candidates for particular procedures or treatments. Mapping customized messages to patients shows that the health system understands a patient’s needs and concerns. Read more about patient pathways here.
Predictive Analytics: How can you identify the customers who have the most pressing needs within those service lines? Using neural networks, deep learning, and millions of data points, you can train models to predict future healthcare utilization. Predictive modeling involves assessing aggregated data to determine the likelihood of certain potential outcomes coming to pass. In healthcare, it can be used as a risk assessment to predict the types of attention or services a consumer or patient is likely to need in the future, based on their current state of health, behaviors, and lifestyle factors. Predictive strategies can also be used to anticipate costs and more effectively manage the health of specific populations. Read more about predictive modeling here.
Behavior Change: How can we speak to those customers with the right message at the right time? Using clustering algorithms and historical data, we create marketing segments based on previous medical encounters, giving you valuable context into a patient's healthcare experiences. With personalization, providers can deliver more value-based care because they're focusing efforts on individuals who are ready and prepared to receive it, moving them through the treatment journey more efficiently. And while this personalization ultimately benefits patients, tailored messaging is also good for hospital marketers, as it leads to higher rates of conversion. Read more about personalization here.
Attribution Modeling: How can we evaluate whether our messaging is working, and improve our next communication? Combining historical data with careful modeling, you can understand more about the interplay between each of your chosen communication channels. Attribution allows you to divide credit for conversions to a specific touchpoint, determining ROI and pinpointing the most successful tactics. This will ultimately help you lower your cost per acquisition (CPA). Read more about attribution here.
These four key areas are the cornerstone of an effective healthcare CRM strategy. You can focus on each of these best practices individually or inclusively, depending on your goals and resources. As with all outreach, it’s important to test and measure your messages and strategies. Data science is not a silver bullet. Currently no machine can replace human judgement for connecting real-world problems to complex analytic solutions, but with the right CRM, hospital systems can leverage data science to significantly improve their marketing and patient engagement efforts.