It’s been hailed as the next “big thing” in healthcare. Predictive analytics promises to aid in every aspect, from creating administrative efficiencies to improving outcomes. But beyond the algorithmic buzz, predictive analytics isn’t just going to help providers make better decisions, according to Castlight Health’s Chief R& D Officer, Jonathan Rende.
Turns out, predictive analytics has the potential to reveal a lot more for patients, including the competitive costs of procedures and how to better utilize insurance benefits provided by employers. Will the use of predictive analytics in this capacity make things more challenging for providers? As patients, what do we stand to gain from what data has yet to delineate? Rende shared his insights on the subject along what his San Francisco-based health information company is doing to deliver on its predictive promise:
Q
Predictive Analytics seems to be the prevailing new promise that’s going to revolutionize healthcare; from improving patient outcomes and reducing costs to creating more personalized care. That’s a lot of change to deliver. Is the technology where it should be to get that kind of job done at this point?
Predictive analytics has already made a massive impact on healthcare. It’s been an instrumental part of informing our product direction. Castlight sits on a huge stack of data – from claims information to eligibility files to search information. This information powers our predictive and rules modeling, and allows us to target specific segments with the information they want, when they want it. It allows us to predict if a particular segment will engage with our product.
You’re right that this is an exciting time, and we are in the very early stages of all the good predictive analytics can do. Because of our rich data sets, we’ve been able to test our predictive models and rules models on our data. These have shown high levels of accuracy, and allow us to demonstrate these models really do work.
Part of the beauty of software is that it gets better as more people use it, as more data sources become available, and as we learn from other industries. But technology is definitely improving healthcare today, and, starting with our transparency product, Castlight Health is at the forefront of this movement.
Q
We really have only begun to scratch the surface with predictive analytics and its capabilities, but how is it already creating a changing climate in healthcare?
As you know our product, Castlight Action, is already leveraging predictive analytics to create this change. We’ve leveraging predictive analytics to create change, but for what? This creates new ways for benefits leaders to communicate with specific segments of their population. For example, with Action benefits teams can: instantly identify opportunities to engage employees who will most likely need care in the near future through advanced predictive models built off medical, pharmacy, Castlight search data, and demographic data; automatically deliver multi-channel personalized campaigns that direct employees to connect with the appropriate benefit, content or vendor program, to identify high quality, lower cost providers, and to make more informed care choices; track engagement and campaign impact on a daily basis, including page views, vendor connections, provider selection, healthcare utilization, and more.
These are powerful developments that will immediately help people achieve better health outcomes and helping reduce the overall cost of healthcare. The technology behind our predictive analytics will only get better, and we will continue to see progress in these two areas.
Q
Companies like Castlight are really breaking new ground when it comes to personalizing healthcare, including Castlight Action, which allows employers to provide their employees with information to drive better decisions about where they consume their healthcare and what benefits are available to improve and maintain their health. How does it all work exactly? Give me a quick snapshot of how Castlight is driving the data down to enable smarter more effective healthcare consumption?
Castlight Action continuously surfaces insights to the benefits leader; segments and targets the relevant population using personalized, multi-channel campaigns with specific behavior change goals; and delivers real-time impact reporting on those campaigns to the benefits leader. The platform enables benefits leaders to unlock the full value of their benefits strategy by bringing the power of data to enable better employee decision making, thus driving lower cost, improved outcomes, and increased benefit satisfaction.
The key to Action is that it segments the population in a predictive manner. It directs information to this population, via multiple channels, when, where, and at the right time the users need it. The timeliness of the information is key – as it is an important part of influencing their behavior. And if we’re able to direct users to the right provider and service at the right time, when they need it most, that is what leads to higher engagement.
In other words, Action leverages all the data on the platform to do a number of things: Action suggests better care options for common but expensive and vital treatments like treatment for back pain early in the patient’s care cycle; similarly Action enables benefits leaders to create compelling content – emails, internal communications – to promote benefits like wellness programs and other care options. Action makes all of this possible, processing the large amount of data on the platform and combining it with a beautiful user interface.
Q
How are you able to personalize the abundance of data and dwindle it down to change those crucial, personal pinpoints that affect the way healthcare services are consumed individually? Seems like quite a feat and a lot of data to drive down. How do you ensure your data is clean and quality enough to produce the results you promise to deliver?
We’ve created a framework to surface and identify different opportunities. These opportunities could be specific conditions (diabetes, lower back pain, etc.) or could be characteristics of a population (non-frequent healthcare shoppers). Castlight helps to turn all users into better healthcare shoppers and decision makers. We look at each opportunity individually. We have amazing data scientists who review these opportunities and prioritize those that have the biggest impact on any population. In Castlight Action, we’re starting with 3 initial opportunities, but each major release after will include additional opportunities with unique characteristics. This way we can better segment those with a specific condition, or specific characteristics.
Q
Can the mining of predictive analytics be trusted at this point from a technological standpoint? Are the capabilities where they should be to really foster the change we are talking about here? For example, how does an employer know a company like Castlight is a better choice than its competitors? What are the distinctions one should look for when it comes to really dialing down on the data?
Similar to my answer to the last question, we look at opportunities one-by-one. We test our own data through our strategic analytics team – which includes some of the leading experts in data science. Through our own data testing, we can develop definitive results by following predictive and rules-based options. And these comprehensive tests happen before any opportunities move into wider development or customer’s hands.
Q
Sure, the information to make the better decisions can now be at the health consumer’s disposal (which will hopefully improve outcomes and reduce costs), but is that really going to influence change required to do those things? Aside from employer-provided incentives, why should health consumers change what they’ve been doing for decades? Is it merely driven by the fact that no one can afford (quite literally) to make mindless decisions anymore, especially with the creation of insurance products like high-deductible health plans? Are there other motivations the data can help derive?
The data and the Castlight approach is all within the technology and platform we offer users. We believe that this will influence behavior change because we’re applying the same approach that all leading SaaS companies with strong engagement are also implementing. We’ve worked with the Stanford Persuasive Technology Lab to apply the BJ Fogg behavioral model, in order to identify motivated users. The same type of modeling is used by companies like Facebook, Google, LinkedIn, to drive engagement on their platforms.
It’s all about getting the right information in front of the right employee at the right time. Our algorithms can pinpoint which employees are ready to engage with their health benefits, and we know they’re in turn more likely to respond well to some of the suggestions Castlight Action provides. Timing is also crucial. For example, if an employee is considering pregnancy, they likely have a multitude of questions about the right doctor and the right appointments to make along the way. A mother-to-be may be unlikely to do any research outside what her doctor tells her. But with Castlight Action, we can ideally have some insight into the fact that this woman is considering pregnancy, and target her with the appropriate information on finding the right OB/GYN and other appointments she’ll need to make. In the long run, this advance knowledge ensures the employee finds the best doctor and avoids any costly mistakes or unnecessary treatments.
Q
Speaking of consumers’ motivations, you have some interesting and astounding data about the disparity in cost of healthcare services/procedures across the country. Will predictive analytics continue to unmask these disparities? Will those cost revelations be another driver to get consumers more involved with taking care of their health and making better decisions about where to receive their care?
Predictive analytics specifically works to segment populations, and ensure that we help people make better healthcare decisions. Our transparency (healthcare price and quality) product is what will unmask the disparity in cost of healthcare services/procedures. Transparency in healthcare price/quality is the backbone of Castlight, and continues to drive healthcare consumerism.
Q
Any insight on what has driven the cost disparity in the country? Was it driven merely from the reality that ignorance is/was bliss when it comes/came to the cost of healthcare for consumers? Can some of the cost differences be justified? Is there more than geography to consider that accounts for those cost differences?
This is a broad issue – particularly as you look at the relationship between payers and providers. There are narrow networks, centers of excellence – and all create independent agreements resulting in different prices for the same medical services. Cost disparity exists because the consumer of healthcare generally hasn’t had a role to play as they would in typical supply and demand markets. They’ve been disintermediated from the process of agreements and price negotiations. But we’ve found with our core Transparency solution that when consumers are engaged in their healthcare cost & quality decisions, they can have an impact on cost.
Q
Is the jig up for healthcare providers when it comes to charging higher than average prices for routine procedures or health services? Will these cost revelations put the higher-priced providers out of business, or merely drive a reduction and leveling of costs?
It’s more of the latter, the drive to reduction and leveling of cost. It’s all about education and transparency. Our goal is to bring more transparency to the industry so consumers can make smarter decisions. They may still choose to visit a higher-priced physician based on factors like reputation or convenience– but we’re hoping to make it clear that there’s no correlation between cost and quality. Castlight’s core focus is on enterprise benefits management, so the immediate goal of increased transparency is to help employees make smarter decisions and save money– but we also want to help reduce waste nationwide. There’s $300 billion wasted in the U.S. every year in avoidable healthcare costs. By providing data on provider quality, offering accurate pricing information and targeting behavior change, we’re working to reduce that waste.
Q
Is it possible that this continued transparency will drive a leveling of healthcare costs nationwide? Or is that unrealistic?
It’s a bit of both. It’s typically first found on the regional basis – and if a region finds itself moving towards a more normalized cost structure, that will have a wider effect across the country.
Q
It seems like predictive analytics is good news for patients and good news for employers, but not so great news for providers. Sure there are analytics that can help them reap the benefits of reimbursement opportunities, but doesn’t this transparent shift put a lot more pressure on providers? And what about payers? Who’s going to be the ultimate benefactor from predictive analytics?
I think providers are more likely to be indifferent to the opportunity segmentation with predictive analytics. As we identify populations and opportunities, it will provide information, and promote services – so provider programs can potentially see more use. But one fuels the other, in that without engagement, consumers can’t make an impact on pricing. Transparency is the backbone of a healthcare engagement platform. Through identification of opportunities on this platform, and engagement with the platform, then prices will go down because consumers will be making educated decisions based on all of the information (including cost/quality/safety) offered to them.
Q
What other possibilities will predictive analytics reveal as it evolves in the healthcare industry? What do you suspect the future of data-driven healthcare looks like?
I believe the future of data-drive healthcare will bring the sharing of data across organizations, delivering a level of visibility to groups like the benefits brokers/consultants to help their clients with information they’ve never seen before. Currently, theses groups don’t have the ability to take action on real-time data. Action can add value to their services, making them trusted advisors to the organizations they’re working with.
Q
There has been a lot of skepticism in the industry when it comes to embracing new information technologies in healthcare. Is power of predictive analytics the ultimate convincing element, do you think? What’s left to deny at this point?
I don’t believe it’s the “ultimate convincing” element, but it’s a necessary building block to make decisions based on data and, one step further, taking action on real data.
Q
We’ve talked a lot about the possibilities, but what if any are the challenges that still lie in the way of predictive analytics and its potential in healthcare? What are technical roadblocks? What are the human road blocks? What needs to happen to get us where the technology can really take us?
One consideration is that there’s always a balancing act between privacy and compliance, versus what you can do with the technology and personal data, and taking action on it. It is a fine line to protect privacy but be able to communicate with users in the right way. Castlight Action was built with all the privacy and security implications at top of mind, and has struck that balance of being compliant but also engaging users to take the appropriate action(s).
Q
Any closing remarks?
We’re only at the beginning of this journey, and it’s incredibly exciting. Particularly with Castlight Action, we’re at the start of identifying the opportunities that will allow benefits professionals to engage with their employees; we’ve already seen that this works and resonates with both employers and employees.
