July 15, 2024

Health Benefit

Healthy is Rich, Today's Best Investment

What Are the Benefits of Predictive Analytics in Healthcare?

6 min read

Predictive analytics in healthcare plays a major role in improving care delivery and patient outcomes. By leveraging historical data, this type of analytics allows health systems to gauge what’s likely to happen in the future, both from an operational and clinical perspective.

This capability is particularly useful for healthcare organizations pursuing value-based care, as the ability to predict outcomes before they happen can help stakeholders identify where their current strategies may be falling short and work toward shifting them.

This is especially relevant for risk stratification and chronic disease management, which, when undertaken effectively, can significantly reduce adverse outcomes and associated costs.

Here, HealthITAnalytics will explore how predictive analytics can benefit patients and providers.


Clinical decision support is one of the most impactful use cases for healthcare predictive analytics.

Successful risk scoring can bolster clinical decision-making, enabling health systems to identify risk factors within a patient population and pursue risk mitigation. Risk scores are developed by flagging relevant risk factors for an adverse event, such as a family history of high blood pressure, and investigating how one or more factors impact a patient’s risk.

Risk scores can then be incorporated into risk-scoring models, which pull data from multiple sources to stratify risk on an individual or population level.

Risk scoring and stratification have many use cases in healthcare, including helping care teams forecast disease progression or treatment success. This is particularly useful for clinical decision support in chronic disease management.

Chronic disease management centers on structuring treatment plans to help patients manage their conditions and improve their quality of life. Predicting treatment efficacy quickly can help clinicians decide whether to alter a patient’s care plan or continue their existing therapy.

However, patients respond differently to different types of treatment, which can present a challenge for clinicians. University of Michigan Rogel Cancer Center researchers sought to address this hurdle for patients with human papillomavirus (HPV)-positive throat cancer by developing a predictive model to forecast whether a treatment method was working months earlier than standard imaging scans.

The research team underscored that currently, clinicians rely on patients receiving these scans every few months to determine if their tumors are shrinking and the treatment is working.

However, this doesn’t necessarily present an accurate assessment because some cancers demonstrate pseudoprogression, a phenomenon in which treatment has been successful, but it can be difficult to tell because a tumor will initially grow before it eventually shrinks.

To mitigate the challenges presented by pseudoprogression, the researchers set out to create a blood test to determine whether a treatment is likely to work after a single cycle.

Using this method of predictive analytics, the blood test could enable medical professionals to assess how a patient responds to treatment months earlier than previously available. This would allow providers to switch their course of treatment sooner if the current one is not working, saving patients months of unnecessary and painful treatment.

Predictive analytics allows healthcare professionals to quickly analyze data and plan a course of treatment that will work best for their patients, saving time and producing better outcomes.


Alongside clinical decision support, predictive analytics plays a pivotal role in population health management.

Using predictive modeling, healthcare stakeholders can track care trends — such as disease prevalence and comorbidities — within a patient population or segments of the patient pool. These data can lend themselves to a host of population health management efforts, like preventing hospital readmissions or encouraging preventive care uptake.

Children’s of Alabama uses a predictive modeling approach to anticipate patient deterioration and extubation readiness in its cardiovascular intensive care unit (ICU). This work supports a larger strategy called ICU Liberation, “an evolved philosophy and practice of improving care by freeing patients from pain, oversedation, delirium, mechanical ventilation, immobility, isolation, sleep disturbances, and ICU-acquired weakness, as well as post-discharge residual effects that can be life-altering for so many patients.”

Predictive analytics may also have an increasing role in the care coordination of populations disproportionately impacted by climate change.

The climate crisis is an ever-present threat to vulnerable populations, and its ramifications for healthcare include increased costs related to illnesses, injuries, doctor and emergency room visits, and premature death, alongside related issues of long work hours and lost wages.

These threats to public health have put a strain on health systems, which often have limited capacity to help people affected by climate change-related disasters like hurricanes and wildfires.

Predictive analytics has helped one healthcare provider expand that capacity.

Umpqua Health, an Oregon-based coordinated care organization (CCO) primarily serving Medicaid beneficiaries, has deployed an analytics-driven population health management platform to identify high-risk members of their patient population who may need emergent care due to wildfire smoke exposure.

The health system pulls electronic health record (EHR) data, social determinants of health (SDOH), patient demographics, language, geography, gender identity, sexual orientation, and other information to perform risk stratification. These insights are then used to flag patients who may benefit from receiving an air purifier ahead of the wildfire season.

From there, care coordinators perform outreach to connect these individuals with air purifiers and other services. As a result of implementing the risk stratification program, the health system has successfully gotten patients to complete health risk assessments and enroll in care coordination programs at higher rates than in the past.

Initiatives like this can also have a positive effect on care quality.


Clinical decision support and population health management contribute to a healthcare organization’s value-based care strategy, even without predictive modeling. But, using predictive analytics effectively can have a positive impact on patient and provider engagement, which has the potential to ease some of the challenges stakeholders face in navigating the transition to value-based care.

Payers and providers have been exploring how to use predictive modeling and other types of analytics to pursue value-based care success.

Elevance Health and MVP Health Care leverage predictive analytics to drive care coordination efforts for their members. Both organizations underscored that care barriers, such as SDOH, keep many patients from undergoing necessary care.

Although educational efforts stressing the importance of routine care can help reduce care barriers, more extensive strategies are needed to accomplish equitable care. To this end, MVP gathers data on its member populations to help predict healthcare needs contributing to adverse outcomes, which the payer then uses to connect members with necessary services.

Elevance utilizes a similar predictive analytics approach to flag high-risk members and outreach to those individuals to help coordinate care.

On the provider side, accountable care organizations (ACOs) are key to shifting toward the value-based care model. Investing in improved care management for patients can help providers improve their efforts, and predictive modeling can provide an even bigger boost.

Buena Vida y Salud ACO partnered with the Health Data Analytics Institute in April 2023 to meet its value-based care goals through predictive modeling and digital twins. In doing so, the ACO aims to “keep patients healthy at home.”

These tools allowed stakeholders to assess unplanned admission risk, worsening heart failure, and pneumonia development. Reports for these populations were then pulled and used to develop risk-based patient cohorts.

When lists of these cohorts and their needs were brought to providers to help guide care management, however, the ACO ran into a significant challenge: care teams were overwhelmed by the amount of data and insights presented to them.

From there, ACO leadership worked with providers to home in on the most actionable insights. This approach was largely successful because it focused on the most important use cases that could be easily implemented into existing provider workloads rather than burdening care teams with additional data and tools that were less useful.

By engaging with providers and letting them choose the patient cohorts or use cases they felt most able to tackle, the ACO no longer had to rely solely on care coordinators to guide the organization’s value-based care work.

These steps — collecting high-quality data, defining the use case, making the predictions actionable, and gaining buy-in from relevant stakeholders — are the basis of a successful predictive analytics strategy in healthcare. By adopting predictive analytics, healthcare organizations can reap major benefits for their patients and providers.


Leave a Reply

Your email address will not be published. Required fields are marked *