Thrive in a changing market with analytics as a core competency

Size: px
Start display at page:

Download "Thrive in a changing market with analytics as a core competency"

Transcription

1 Thrive in a changing market with analytics as a core competency Data and new approaches to analytics are a target for investment for every healthcare organization, according to Healthcare IT Trends for 2018 from USF Health Online. It is a core tenant of value-based care and the driver to reduce costs while improving outcomes, member engagement and will play a major role in healthcare innovation and transformation as we move through 2018, the authors wrote. Healthcare organizations that succeed in the rapidly evolving market will have analytics as a core competency. This success will be gated by the barriers of affordability, flexibility and access to skilled resources. 1 This article examines three analytics-enabled solutions from NTT DATA Services that health plans can use to drive new business growth and better manage key business indicators such as medical costs, population risk and member engagement rates. What NTT DATA is doing is bringing solutions to our customers that are innovative, multi-purpose and help them address a range of operational barriers to using next-generation predictive analytics, says Dr. Suman De, Global Solutions Lead for Health Plan Analytics at NTT DATA.

2 Cognitive engine-driven predictive analytics The creation of the citizen data scientist Health plans need specialized predictive analytic solutions to improve care, reduce costs and enhance member satisfaction. But they face two problems in doing this: predictive solutions are expensive and time-consuming to implement, and running them typically requires data scientists with healthcare experience, who are difficult to find. Augmented analytics is a particularly strategic growing area which uses machine learning to automate data preparation, insight discovery and insight sharing for a broad range of business users, operational workers and citizen data scientists, said David Cearley, vice president and Gartner Fellow. Mid-market and regional health plans have an aspiration to use data and analytics technology to serve members and grow, says Karen Way, MHA, Global Solutions Lead for Health Plan Analytics at NTT DATA. But many don t have the ability to attract and retain the resources required. This is true from both a cost perspective as well as availability; data scientists with healthcare experience are not easy to find. To overcome these hurdles, NTT DATA partnered with a data science technology company to create a solution that addresses both problems. It allows healthcare organizations to build and develop AI-driven predictive models, with or without data scientists. It allows business subject matter experts to become citizen data scientists, working directly with the tool to develop their own algorithms. This approach makes it easier and faster for organizations to develop and manage predictive models. Unlike other software solutions, NTT 2

3 DATA s cognitive engine provides complete visibility into the predictive model s development, and users can manually configure or modify the models as needed. Users get the flexibility to run their own algorithms (developed outside the engine) alongside those the engine recommends. Organizations can design, implement and begin generating insights from the organizational data assets in fewer than three business days. Health plans may have a lot of people who know data well but are not data scientists, says Dr. De. Technology that doesn t require a data scientist to run and makes advanced analytics accessible and doable for these employees is a huge value for health plans. Advanced analytics tools also support professional data scientists by helping them work more efficiently. Automating the laborintensive task of developing complex algorithms for predictive models frees up data scientists time, allowing them to work more closely with the business experts. These tools do the heavy lifting, so these scientists can focus their full attention on more complex problems and in-depth research says Dr. De. Predictive analytics driven by a cognitive engine can ingest, interpret, analyze and act upon data while continually building knowledge through unsupervised learning. The cognitive engine automates and simplifies building and implementing complex predictive models, e.g., Neural Network, Random Forest, Naïve Bayes, Regression, etc. This enables organizations to deploy and obtain actionable insights with increased speed. Cognitive engine at work in healthcare A large regional health plan partnered with NTT DATA to implement its cognitive solution. The There are no restrictions on the type of data the cognitive engine can handle. For this use case, the engine consumed structured and unstructured data, including claims data, authorization data, lab data, health risk assessments, member historical risk scores, care management notes, member call services notes, demography and enrollment data, and even socioeconomic data. Dr. Suman De, Global Solutions Lead for Health Plan Analytics, NTT DATA 3

4 It s not so much about going after and recovering 100 percent of improper payments as it is about identifying the root cause of error in your claims processing. Is it a CPT code that s no longer valid? Is it a processing rule that s now outdated? When you know the cause, you stop it from occurring rather than going through the timeconsuming and costly process of cost recovery. Karen Way, MHA, Global Solutions Lead for Health Plan Analytics, NTT DATA organization wanted to identify members who were likely to move to a higher risk category in the near future. Early intervention by a medical management team can often make a difference in the outcomes and costs of these rising risk patients by ensuring they get the right care. The health plan ran 24 months of retrospective claims data, comprising one-third of its total member population, through the cognitive engine to train the engine and create the predictive model. Then it analyzed current data using the model created. This was accomplished in just three business days, a dramatically shortened timeframe compared with traditional approaches. Within the segment analyzed, the engine identified 30 percent more members with potential to move into the high-risk category than the analytics model they had been using. Insights obtained for the high-risk members made it easier for medical management teams to determine the appropriate programs and interventions, potentially saving millions of dollars for the health plan. There are no restrictions on the type of data the cognitive engine can handle, says Dr. De. For this use case, the engine consumed structured and unstructured data, including claims data, authorization data, lab data, health risk assessments, member historical risk scores, care management notes, member call services notes, demography and enrollment data, and even socio-economic data. This flexibility means health plans can feed data through the engine to obtain insights into a range of clinical and financial business needs, such as identifying potential gaps in medical care, detecting claims that were incorrectly paid, and more. Medical cost management solution Identifying the root cause of claims errors The use of predictive analytics is one way health plans are working to reduce medical spend and drive business growth. Another is ensuring claims payment integrity, which plays a key role in helping insurers contain costs. 4

5 With the advent of value-based care, health plans are under extreme pressure to manage their spend and keep their medical loss ratio near or at 80 percent, Ms. Way says. Our medical cost management solution does a total claims analysis to help health plans understand the root cause of why the medical costs are what they are. Unlike other payment integrity solutions, NTT DATA s medical cost management solution is a rules-based engine that analyzes 100 percent of health plan claims instead of just a portion of claims. The result is a much better understanding of adjudication and payment processes across the organization. It s not so much about going after and recovering 100 percent of improper payments as it is about identifying the root cause of error in your claims processing, Ms. Way says. Is it a CPT code that s no longer valid? Is it a processing rule that s now outdated? When you know the cause, you stop it from occurring rather than going through the time-consuming and costly process of cost recovery. Gaining visibility into claims adjudication processes is particularly important for health plans engaged in value-based, bundled or capitated contracts with providers, where errors can be very expensive and labor-intensive to resolve. Ensuring bundled and capitated payments are paid appropriately is increasingly important in the transition to value-based purchasing, Ms. Way says. Consider one health plan that used NTT DATA s medical cost management solution to improve payment integrity. The analysis found an instance where a bundled payment became unbundled during the adjudication process; the health plan paid each line item rather than the lump sum 5

6 allowed. The result was an improper payment of $11,000 to the healthcare provider instead of $300. That was just one of the cost recovery opportunities they found; overall, they identified nearly $3 million in excess payments within four months of claims data. With the ability to retroactively review claims up to 24 months, the health plan could recover a significant amount of revenue. In addition to capturing cost recovery opportunities, the medical cost management solution also helps identify providers who are not performing to the standards required by their value-based contracts. Health plans can use the knowledge gained through this analysis to work with providers to improve performance, guide members to betterperforming providers or to remove substandard providers from their networks. Sentiment Analysis Platform How satisfied are my customers? Health plans currently track their member satisfaction rates via the Consumer Assessment of Healthcare Providers and Systems (CAHPS) surveys. This survey is used by CMS as an input to a health plan s Medicare Advantage (MA) Star ratings, which influence the reimbursement rate for Medicare claims. Health plans have limited insight into the responses on these surveys until the Star ratings are published. Maintaining high levels of member satisfaction is key to health plans strategic and financial success. Health plans with MA Star ratings of 4.5 or greater receive better reimbursements from CMS for MA claims. Health plans with high scores can also enroll new plan members throughout the year, easing the need to compete with other health plans to capture their entire patient population during the open enrollment period. Organizations with MA plans that score 3 stars or below for three consecutive years run the risk of CMS barring them from offering MA plans altogether. Natural Language Processing (NLP) is a vein of data science that draws value from and identifies patterns in large sets of unstructured or written data. Applied to online and social media content, it can help health plans identify member sentiments and behavior and address their concerns in real time. NTT DATA has a solution that allows health plans to use NLP to capture member sentiment using online and social media data. This provides an opportunity to address issues and improve the member experience proactively, rather than after survey results affect their ratings. There is a wealth of healthcare-related information on social media. Plan members use social media channels to vent about a poor physician interaction, solicit donations and financial aid for medical bills, and query friends and family for healthcare recommendations. Healthcare was the third most-reviewed type of service on Yelp in 2017, according to a consumer survey done by Bright Local. Online health reviews are not only more voluminous in scope than CAHPS data, but social media channels also provide valuable insights health plans can t get from traditional patient and customer surveys. For one, online patient testimonials are continuously updated and reflect changing sentiments and experiences in real time, unlike static and less time-sensitive measuring devices like CAHPS. Furthermore, because social media reviews are unstructured, they capture a greater depth of information. For example, Yelp reviews cover an additional 12 domains not reflected in CAHPS, according to a study in Health Affairs. NTT DATA s social listening and sentiment analysis platform uses advanced data mining concepts and NLP to scour and analyze social media activity and predict member sentiments about a health plan s brand. The solution 6

7 transforms large volumes of unstructured online data and member comments into actionable information, highlighting opportunities for health plans to implement changes, better influence their population and increase engagement with members in near real time. The tool enables health plans to identify and address the root causes of the low performing customer service measures that lead to poor Star ratings and improve their eligibility for increased CMS financial earnings. Conclusion Analytics is quickly becoming a core competency for health plans of all sizes. Advancing data science technology is helping eliminate barriers to using advanced analytics capabilities for organizations, from affordability and upfront costs to skilled resources. Cognitive enginedriven predictive analytics solutions, medical cost management solutions, and member sentiment analysis platforms empower health plans to gain insights into their member risk patterns, medical spend trends and financial operations faster, cheaper and easier. All three of NTT DATA s solutions are affordable and scalable. Another benefit: all can be implemented and executed within short timeframes, thus enabling the health plan to be more nimble in responding to market changes. Health plans don t have to choose between technology that produces high quality results, and technology that deploys quickly and at a lower cost, Dr. Suman says. Quality, cost and efficiency are no longer mutually exclusive. Health plans don t have to choose between technology that produces high quality results, and technology that deploys quickly and at a lower cost. Quality, cost and efficiency are no longer mutually exclusive. Dr. Suman De, Global Solutions Lead for Health Plan Analytics, NTT DATA Learn more at BECKER S HOSPITAL REVIEW 7