Analytics is typically used to make sense of data. Traditionally, it results in an understanding of the past. Predictive analytics is turning this equation on its head by moving beyond retrospective analysis and providing insights on what is likely to happen in the future. Such capabilities could change the game for healthcare payers and healthcare systems – especially when it comes to risk adjustment initiatives.
An understanding of predictive analytics, how it differs from retrospective analytics, its risk-adjustment use cases, and what healthcare organization (HCO) leaders should look for in a technology partner can help healthcare payers and systems experience the outcomes that will lead to success not only today but the days, months and years ahead as well.
Defining predictive analytics
Predictive analytics leverages data mining, modeling, and machine learning to analyze current and historical facts to determine what will happen in the future.1 More specifically, predictive analytics involves applying statistical analysis techniques, data queries, and machine learning algorithms to data sets to create predictive models that place a numerical value – or a score – on the likelihood of a particular action or event happening. As such, predictive analytics has become a promising discipline in the field of data analytics, an umbrella term for the use of quantitative methods and expert knowledge to derive meaning from data and answer fundamental questions.
Because predictive analytics is so powerful, it is becoming increasingly commonplace. Consider the following: According to a study by Allied Market Research, the cumulative average annual growth of the predictive analytics market is expected to hit 21.9% from 2020 to 2027 and the value of the predictive analytics market will reach $35.45 billion by 2027.2 Healthcare is certainly part of this growth. According to a report from market research company IMARC, the global healthcare predictive analytics market will grow from $6.32 billion in 2021 to $22.59 billion in 2027.3
Predictive analytics and risk adjustment
Risk adjustment is tailormade for predictive analytics. Simply consider the following: According to HealthCare.gov, risk adjustment is “a statistical process that takes into account the underlying health status and health spending of the enrollees in an insurance plan when looking at their healthcare outcomes or healthcare costs.”4 The Centers for Medicare and Medicaid Services (CMS) takes the definition one step further, explaining that risk adjustment models use actuarial tools to predict healthcare costs based on the relative actuarial risk of enrollees in covered plans.5
Here’s how it works: In a risk adjustment model, a patient gets a risk score based on demographics, such as age and gender, as well as health status. ICD-10-CM codes, which represent a patient’s diagnoses, provide data about health status and, therefore, the expected outcomes and costs of care. ICD-10-CM codes are linked to hierarchical condition categories (HCCs), a set of medical diagnoses that each have an assigned risk adjustment factor (RAF). RAF scores are reflective of the risks associated with the patients.
This coding process empowers healthcare organizations (HCOs) to make proper documentation and reporting of diagnoses essential to the success of risk adjustment programs.6 This is especially important for healthcare payers and systems that work under capitated risk adjusted models, where they are reimbursed a set price for each member.
Predictive analytics is so important because it can help these HCOs succeed under such scenarios by making it possible to see the future, then take action to provide optimal care, contain costs – and ultimately maximize margins.
For example, with predictive analytics, HCOs can anticipate a member’s likelihood of certain conditions developing in the future. To do this, digital predictive analytic tools can scrub through clinical content such as medical records and unearth conditions that previously were not associated with the patient. As such, predictive analytic tools could assess family history and other medical indicators and determine that a patient has or is close to developing diabetes or chronic obstructive pulmonary disorder (COPD), two conditions that often go unnoticed.
Predictive analytics: risk adjustment applications
While predictive analytics helps to bolster risk adjustment overall, there are specific applications that healthcare leaders should consider:
Clinical suspecting identifies members with suspected but potentially undocumented conditions. By leveraging large data sets and advanced queries, clinical suspecting identifies year-over-year trends and historical gaps in a member’s chart. This can include missing procedures, specialist visits, or previously documented conditions. These analytics also capture the opportunity for new diagnoses that may apply to risk-adjusted conditions.
Suspect analytics provide insight on a greater healthcare picture. Implementation of suspect analytics enables identifying members with potentially unrecorded or undocumented conditions. This technique also helps in identifying significant gaps in care by improving the accuracy of risk models. The predictive analytics methods run through volumes of historic and current data to determine patterns of specific conditions and determine the probability of occurrences in future. Suspect analytics enables significantly optimized risk-adjustment revenue.
Predictive analytics leverages data mining, modeling, and machine learning to analyze current and historical facts to determine what will happen in the future.
By using these suspecting techniques, HCOs can more accurately assign the right resources to various conditions. When risk is captured accurately, HCOs can dedicate the right time and financial resources to certain conditions in an effort to cost-efficiently achieve optimal outcomes. Being able to accomplish this in a predictive manner empowers HCOs to proactively take the steps to match resources to conditions – and to, therefore, better meet patient needs and achieve the outcomes necessary to succeed under risk-based models.
With predictive analytics, it is possible to identify the most vulnerable patients in a HCO’s population – and then prioritize outreach to these patients. By leveraging extensive data sets and advanced analytics knowledge, an HCO can build a flexible scoring model to risk stratify its entire population. In essence, such models make it possible to predict and identify at-risk vulnerable populations, and then prioritize which patients would benefit from care. EXL, for example, helped a large healthcare provider organization stratify its patient population in low, medium, and high vulnerability during the COVID-19 pandemic. At the peak of COVID uncertainty, the provider was able to schedule more than 80% of its vulnerable patients to have consults or in-person visits, a rate higher than pre-COVID measures.
Improved performance ratings
The Healthcare Effectiveness Data and Information Set (HEDIS) is a group of standardized measurements used to evaluate provider performance in terms of clinical quality and customer service. Developed by the National Committee for Quality Assurance (NCQA), HEDIS data is a retrospective review of healthcare services to ensure clinical care is meeting quality standards and that providers are following evidence-based guidelines. Health plans and other health care entities collect these key measurements and submit the findings to NCQA, resulting in an aggregate quality comparison across health plans.
Because predictive analytics can provide healthcare leaders with insight into future performance, HCOs can anticipate needs and take action to improve outcomes in the six domains of care that HEDIS targets: effectiveness of care, access/availability of care; experience of care; utilization; and measures collected using electronic clinical data collection.
With optimal performance in each of these areas, members are likely to be satisfied with the service they are receiving and will be more apt to provide positive reviews. Such feedback from members translates into high HEDIS Star Ratings, a rating scheme created by CMS.
Predictive analytics can also provide the insights that help HCOs optimally manage members’ chronic conditions. For example, if certain test results or vital signs are heading in a particular direction, the HCO can take action and proactively outreach and care for the member. In addition, HCOs can predict that certain members are moving toward developing a condition and can proactively create an intervention plan. By getting out in front of conditions, HCOs can achieve the best outcomes, while also reducing costs.
Targeting the right cohorts
Predictive analytics can help HCOs target patient or member cohorts that are most amenable to medical interventions or change. By using predictive analytics tools, HCOs can not only identify high-risk members but also understand which of these members are most likely to experience the greatest return on investment from various future care interventions. In addition, such analytics make it possible to identify which care gaps are easy to close. Perhaps most importantly, however, predictive analytics can also identify which members will have the highest propensity to engage with suggested medical care plans and interventions in the future.
Predictive analytics can bolster documentation efforts as well. Such analysis, for example, can help to identify conditions that have been documented year after year with a high degree of accuracy and precision – and point the healthcare professional toward once again documenting these conditions. In essence, the tools can produce a confidence score that indicates how likely it is for a certain condition to require documentation each year.
Predictive analytics can also shed light on a HCO’s documentation trends and abilities. Such scorecards can indicate that a particular provider is very adept at documenting certain conditions but fails to document others. As such, the HCO can work with clinicians to encourage improved documentation practices in the future. In addition, predictive analytics can apply artificial intelligence to EMR data to determine where clinicians are providing care and are not adequately documenting for their services. Such analysis can prompt providers to prioritize patients in the future by ensuring that an evaluation is conducted, the condition is verified, treatment is initiated – and documentation occurs.
Predictive analytics delivers future insights and actionable information to help provide optimal care, contain costs – and ultimately maximize margins.
When seeking to implement predictive analytics for risk adjustment purposes, HCO leaders need to enter into partnerships that can move their efforts forward.
EXL Health is in a position to help.
EXL Health has leveraged considerable experience, extensive data sets and advanced analytics knowledge to build powerful risk analysis solutions that employ predictive analytics.
For example, EXLCLARITY™ improves value-based care strategies with a holistic view of patient-population risk. The platform, which earned the prestigious Best in KLAS distinction, aggregates claims, clinical and other data from CMS, private insurers and client electronic medical records to deliver detailed predictive analytics, natural language processing/artificial intelligence-powered chart reviews, and near-real-time views of member/patient risk scores.
The EXL Difference
HCOs can confidently leverage predictive analytics to advance risk adjustment initiatives and improve outcomes because EXL brings the following to the table:
EXL relies on significant healthcare, data analysis and risk adjustment experience to create all of its solutions. EXLCLARITY™ is one of the earliest tech platforms that’s focused on the risk adjustment use case.
A reliable truth
It’s important for HCOs to access a single source of truth for all risk adjustment purposes and all risk adjustment scores associated with a patient. The EXL platform accepts and normalizes all incoming information from multiple sources, keep track of the incoming data and associating relevant information with particular patients. As a result, HCO professionals move forward with confidence, knowing that they are working with the most accurate risk score and zeroing in on the most pertinent gaps from a clinical perspective.
EXL Health leverages considerable experience, extensive data sets and advanced analytics knowledge to build powerful risk analysis solutions.
Multiple sources of data
While EXLCLARITY™ produces one source of truth, such insight is derived from multiple sources of data including claims and clinical content. EXLCLARITY™ even brings social determinants of health (SDOH) data into the mix. With information on where members live, learn, work and play, it is possible to better meet their healthcare needs7. The ability to leverage SDOH empowers HCOs to implement CMS codes that account for variables such as transportation issues.
Because EXLCLARITY™ works with a wide variety of data, it is possible for HCOs to calculate – and subsequently act upon – accurate risk scores.
One view. EXL presents insights in one dashboard, making it easy for HCO leaders and staff to access, understand and take action on various insights.
Various advanced technologies can help HCOs better manage the predictive analytics process. EXLCLARITY™ relies upon a workflow module to help streamline various processes and natural language processing (NLP) to meet numerous reporting requirements.
A reusable yet flexible model
By leveraging its experience with multiple clients and various types of data, EXL has created a model that is replicable across the board. While EXL does not build a new solution for each HCO, it does offer the flexibility in how HCOs take in and ultimately leverage various insights to bolster risk adjustment initiatives and improve outcomes. For example, one health plan might want to analyze the risk associated with 80% of its members, while another might just analyze the 20% that represent the greatest amount of risk.
The ability to work hand-in-hand
EXL works as a collaborative partner with each HCO. Instead of taking a one-sizefits all approach, EXL works with HCOs to review their needs and challenges and then uniquely implements its technology to help eacn organization move forward. In the 2022 Best in KLAS - Risk Adjustment report, EXL earned an A+ score in the relationship category, underscoring the company’s commitment to a deep understanding of customers’ businesses and its focus on building collaborative working relationships.
While EXL provides powerful technology, the company also ensures that HCOs remain engaged and continually leverage the data in the right manner. Instead of simply offering insights, EXL offer the realworld guidance that helps HCOs achieve their goals.
For example, predictive analytics can provide call center representatives or care coordinators to not only address a member’s current concerns but to improve the overall customer experience by offering suggested “next best actions.” Such recommendations provide members with the guidance that will help them maintain and improve their health in the days ahead.
About EXL Health
At EXL Health, we provide basic, intermediate, and advanced options to help you succeed.
EXL Health’s modular, end-to-end risk adjustment and quality management solution is a combination of products and services to drive improvements across your quality management ecosystem. Our end-to-end and AAPC-certified risk adjustment and quality services deliver optimized performance on a risk-adjusted population, including improvements across clinical documentation, quality outcomes and accurate reimbursement.
At EXL, our goal is to work with you. We work together with our clients to collaborate and develop lasting relationships based on their specific goals and needs. Do you want to improve your score? Do you want to streamline risk adjustment reviews and HEDIS data collection? Do you have the right resources and tools? Once we understand your unique business challenges, we will customize our solutions to best fit your organization.
EXL Health partners with and delivers our clinical services to both health plan payer and provider organizations. We support driving value-based outcomes and transformation in order to support improvement to member health outcomes, medical cost containment and quality compliance.
- Expert.ai. What is the difference between data mining and predictive analysis? https://www.expert.ai/blog/data-mining-predictive-analytics-difference/
- Allied Market Research. Predictive analytics market to reach #35,45 billion by 2027. https://www.alliedmarketresearch.com/press-release/predictive-analytics-market.html
- Imarc. Healthcare predictive analytics market. https://www.imarcgroup.com/healthcare-predictive-analytics-market
- Healthcare.gov. Risk adjustment. https://www.healthcare.gov/glossary/risk-adjustment/
- Centers for Medicare and Medicaid Services. Risk adjustment methodology overview. https://www.cms.gov/CCIIO/Resources/Presentations/Downloads/hie-risk-adjustment-methodology.pdf
- AAPC. Risk adjustment documentation and diagnosis coding. https://www.aapc.com/risk-adjustment/documentation-coding.aspx
- Incorporating machine learning and social determinants of health indicators into prospective risk adjustment for health plan payments - PubMed (nih.gov)
Vice President, Product and Strategy
Vice President, Healthcare Data and Analytics