How to use predictive analytics in healthcare

Using predictive analytics to contribute to intelligent interoperability

Every minute physicians spend navigating administrative tasks or deciphering scattered data is a minute lost to patient care. Predictive analytics working within AI-native electronic health record (EHR) systems offers a promising solution: streamlining workflows, surfacing actionable insights, and enabling more personalized, timely interventions. When integrated into cloud-native platforms like athenahealth, predictive tools can enhance patient care, reduce clinician burden, and optimize practice efficiency.

AI-native EHRs backed by SaaS architecture serve as an intelligent bridge between the immense amounts of data available across the athenahealth network and the specific workflows providers interact with across the platform. That can help provide more intelligent interoperability, with predictive models within the AI-native EHR helping clinicians get time back for patient care and provide recommendations for enhanced clinical decision making.

How do predictive analytics work in healthcare?

Predictive analysis helps address data overload by using machine learning to help structure data and then learn from that data to identify patterns and probabilities. The AI-native intelligence layer helps sort through a rich foundation of data and synthesize it for clinician use in decision making and care plans.

AI-native EHRs that help structure and analyze data from across the healthcare ecosystem can help clinicians use predictive analytics to their advantage. With more usable data, clinicians can better forecast potential patient needs, risk factors, and existing care gaps. That, in tandem with predictive analytics to optimize operational efficiency, helps everyone in the practice – from physicians to administrative staff – focus on the high-priority items that contribute to better patient outcomes and a stronger patient experience.

By surfacing real-time updates, providing smart automation, and reducing the number of both inbox tasks and cognitive load across multiple workflows, predictive models can help free up additional time for physicians to practice medicine.

What clinical workflows can use predictive analytics?

Structured data can give physicians and all healthcare providers more clarity in how they utilize shared information to make recommendations. This takes place across multiple workflows.

• Assisting with diagnoses and generating care plans, subject to provider review, to help improve patient encounters
• Offloading administrative tasks to help free up additional time for practice administrators
• Scrubbing claims with updated rules to help catch errors and strengthen revenue cycle management

Let’s take a closer look at how predictive analytics work across each clinical, practice management, revenue cycle and patient engagement workflows.

Risk stratification and diagnosis support

By structuring data within the EHR, the AI-native intelligence layer enables clinicians to more effectively use information uncovered in patient charts to identify patients at higher risk and prioritize care interventions.

AI-native athenaOne® will soon be able to use generative AI to help surface potential Hierarchical Condition Category (HCC) diagnosis gaps. For example, a patient may have an imaging result stating they have Congestive Heart Failure, but there is no corresponding diagnosis for CHF in their chart. The generative AI could help flag this for providers, linking to the source and enabling physicians to make a more informed decision about diagnoses. By helping to uncover potential care and diagnosis gaps, the AI-native intelligence layer in athenaOne can help practices provide more comprehensive, up-to-date care and satisfy value-based care (VBC) requirements.

Predictive models can also help identify at-risk patients based on quality measures and other value-based care outcomes, processing vast amounts of shared healthcare data, history, and using actuarial models to help create cohorts. More advanced AI agents could potentially then create outreach campaigns to help them obtain and adhere to care.

Precision medicine

Other predictive analytics tools can be added onto AI-powered EHRs to assign polygenic risk scores (PRS) that estimate a person’s genetic risk for developing a particular trait or disease. OpenDNA is a precision medicine AI platform that combines PRS with clinical data from athenaClinicals® (athenahealth’s EHR system) to help predict a patient’s risk for conditions like cardiometabolic diseases, cancer, and support clinical decision-making. When physicians use OpenDNA within athenaOne, they have access to a patient dashboard that shows how a patient’s risk level could change based on clinical parameters and lifestyle changes, helping physicians understand and apply the predictive insights more effectively.

Condition forecasting

The athenahealth Flu Dashboard provides historical week-by-week data on patients diagnosed with flu-like symptoms nationwide. Clinicians can use trends in the data to anticipate potential future care demands. When flu activity increases, especially in their local area, clinicians can proactively reach out to at-risk patients and suggest they schedule an appointment or offer preventive care guidance.

Automated document labeling

Sifting through external documents and faxes is a tedious job for practice administrators. Fortunately, predictive tools can also help streamline administrative work.

One example is Predicted Document Labels for Admin Documents, a feature athenahealth developed in response to customer feedback. This tool harnesses AI to accelerate the chore of labeling administrative documents by suggesting labels for external faxes. As users interact with the tool and accept or correct predictions, the AI learns and improves its accuracy over time. By allowing clinicians to expedite a repetitive and often cumbersome manual task, this predictive model helps boost clinical efficiency, reduce staff workload, and free up time to focus on more meaningful patient-centered work or higher-value skills.

Scheduling optimization and task automation

Predictive analysis can also assist in resource allocation. By analyzing historical trends, AI models can help clinicians forecast patient no-shows and peak visit times to help optimize provider schedules and reduce downtime. That, in turn, carves out more time for patients and can help strengthen patient engagement. Within the AI-native EHR, predictive models also streamline workflows by automating routine inbox tasks with smart alerts and follow-up reminders.

By surfacing real-time updates, providing smart automation, and reducing the number of both inbox tasks and cognitive load across multiple workflows, predictive models can help free up additional time for physicians to practice medicine.

Deploying predictive analysis to help boost RCM

Timely reimbursement is essential for scalable growth and strong patient retention. Predictive tools embedded within the intelligent, AI-native EHR can support every phase of revenue cycle management, so practices maximize reimbursement and maintain steady cashflow with up to 50-70% less work¹.

Claim scrubbing and denial prevention

When it comes to payer-provider relationships, machine learning can analyze billing and coding documentation and compare it against historical claims data and coding patterns. The rules engine within the AI-native layer of athenaOne’s RCM workflow analyzes billing and coding documents in real time to help predict which claims may be more likely to be denied and suggest actions billers can take to improve acceptance. It can help flag potential errors and enable adaptive claim editing before claims are submitted.

Dynamic claim management

Predictive models can adapt quickly using network data based on payer behaviors, including dynamic claim alarms and anomaly detections that help reduce the chance of denials based on new information and patterns.

EOB digitization and automation

Predictive tools in the AI-native layer can help streamline other areas of the revenue cycle process – from Explanation of Benefits digitization, to prefilling prior authorization forms, to detecting potential underpayments and contract gaps. Having AI tools assist with these tasks can help practices get paid faster and lift the overall bottom line.

How predictive analytics benefit patients

While predictive analytics supports clinicians with risk stratification and diagnoses, as well as creating more personalized treatment plans, it also plays a key role in improving overall patient experience. As practices are more able to anticipate needs and streamline care, care feels faster, more proactive, and more coordinated, and patients feel more satisfied and loyal to their providers.

Patient empowerment

Health insights generated by predictive models can empower patients with more information about their own health risks and outcomes. That gives them more opportunities to engage proactively with their care team by asking questions, adjusting care plans and considering lifestyle changes. By increasing patient involvement, predictive models can help unlock deeper patient engagement.

Importance of collaboration and oversight in predictive analysis

Predictive analytics in healthcare relies heavily on historical patient data and industry trends. But the insights it gives clinicians are only valuable if the data is accurate and up to date.

Human oversight is essential, not only to validate the quality of the data but also to help clinicians understand the limitations and potential biases of AI models. Broader, higher-quality datasets can improve how predictive models are trained, leading to more reliable and actionable insights.

Better data inputs are made possible through continuous interoperability with a steady flow of new shared data. Establishing quality assurance teams to audit and verify AI outputs may also help validate accuracy for clinical insights.

Communication across channels – between payers, providers, and vendors – is critical to ensure accurate AI predictive models. One approach is to pilot predictive analytics in a narrow, controlled part of the organization, allowing teams to refine processes and minimize any risks before broader implementation.

What is the future of predictive analytics?

The use of predictive analytics in healthcare continues to expand. One recently developed predictive model aims to identify high-risk patients for targeted gastric cancer (GC) screenings². Ultimately, it’s possible that predictive analytics models can help anticipate a wide range of patient health risks at very large scales. But this is dependent on the availability and usability of data in health IT systems.

At athenahealth, we want to help clinicians maximize positive patient outcomes. We constantly look for ways to improve our solutions by asking for customer feedback and any existing pain points for clinicians in how they utilize predictive analytics.  That’s why we’re committed to the vision of a cloud- and AI-native EHR leveraging data from providers in the athenahealth network and seamless intelligent interoperability across the healthcare ecosystem.