Joseph Papin, MD on Why Predictive Analytics Should Drive Preventive Care, Not Just Reporting

Healthcare organizations have access to more data than ever before. Electronic health records, claims databases, laboratory systems, pharmacy records, remote monitoring platforms, and patient-generated information can provide detailed views of how patients interact with the healthcare system.

Yet having more data does not necessarily mean acting earlier.

Many organizations continue to use analytics primarily to explain what has already happened: which patients were hospitalized, where costs increased, which quality measures were missed, and how utilization changed over time.

For Joseph Papin, MD, physician executive and Principal of Suncoast Search Capital, the greater opportunity is to use predictive analytics to identify risk before an adverse event occurs and connect that information to preventive interventions.

The value of predictive analytics is therefore not simply the ability to produce a more sophisticated report. Its value lies in whether the resulting insight changes what healthcare teams do next.

From Retrospective Reporting to Predictive Intervention

Traditional healthcare reporting is largely retrospective.

Organizations review hospitalization rates, emergency department utilization, readmissions, medication adherence, quality measures, and financial performance after events have occurred.

These reports remain important for accountability and performance management.

Predictive analytics addresses a different question:

What is likely to happen next, and what can the organization do about it?

For example, an analytics model may identify patients with characteristics associated with elevated risk of hospitalization. That information can then support targeted interventions such as medication review, primary care follow-up, care management, or additional monitoring when clinically appropriate.

This represents a shift from measuring outcomes to managing risk.

For Dr. Papin, that transition is particularly relevant to organizations participating in value-based care arrangements, where preventing avoidable deterioration can have both clinical and financial implications.

Prediction Is Not Prevention

An important distinction is often overlooked.

A predictive model does not prevent an adverse event.

It identifies a probability.

Prevention occurs only when healthcare professionals interpret that information and take an appropriate action.

Consider a patient identified as having a high predicted risk of hospitalization.

The prediction itself does not tell a physician exactly what should happen next.

The patient may have an unresolved medication issue, an untreated behavioral health condition, inadequate follow-up, limited access to transportation, or a clinical problem that requires specialist evaluation.

This is why predictive analytics must be integrated with clinical workflows.

For Dr. Papin, the operational question is not simply whether an organization has an accurate model. It is whether the organization has the people, processes, and governance required to respond to the model’s findings.

Risk Stratification Can Focus Limited Resources

Healthcare organizations rarely have unlimited care-management resources.

A population health team may be responsible for thousands of patients while having capacity to provide intensive intervention to only a smaller subset.

Predictive analytics can help prioritize those resources.

Risk models can incorporate combinations of factors such as:

• Prior healthcare utilization
• Diagnoses and chronic conditions
• Medication patterns
• Laboratory results
• Demographic information
• Claims activity
• Previous care gaps
• Social and behavioral factors, where appropriate data are available

The goal is not to replace clinical assessment.

It is to help identify where additional attention may have the greatest potential value.

CMS and other healthcare programs increasingly use risk adjustment and predictive methodologies to support population-level management, particularly in accountable care and other value-based models.

Clinical and Claims Data Need to Work Together

Predictive analytics becomes more useful when organizations can combine different types of information.

Clinical data can provide information about diagnoses, laboratory findings, medications, encounters, and documented care plans.

Claims data can reveal patterns across healthcare organizations, including emergency department utilization, hospital admissions, specialist services, and prescription activity that may not be visible within a single provider’s electronic health record.

Each source has limitations.

Combining them can create a more comprehensive longitudinal picture of patient utilization and clinical needs.

For Dr. Joseph Papin, this integration is an important component of healthcare operational strategy.

The objective is not to accumulate data simply because it is available. It is to connect information sources that can help clinicians and care teams understand patient risk more effectively.

Preventive Care Requires Timely Information

Preventive care often fails for relatively ordinary reasons.

A patient may miss a screening appointment.

A chronic condition may not be adequately monitored.

A recommended follow-up may never occur.

A care gap may remain invisible because information is distributed across different systems.

Predictive analytics can help identify populations where these patterns are occurring.

For example, a health system could use analytics to identify patients who are overdue for evidence-based preventive services or who demonstrate clinical indicators associated with deterioration.

The intervention still requires human judgment.

But the analytics system can help ensure that the right patients are brought to the attention of the right care team.

Predictive Analytics Can Support Chronic Disease Management

Chronic diseases account for a substantial share of healthcare utilization and spending in the United States.

Conditions such as diabetes, cardiovascular disease, chronic kidney disease, and chronic respiratory disease often require ongoing monitoring rather than isolated episodes of treatment.

Predictive analytics can support these programs by identifying patients whose clinical or utilization patterns suggest increasing risk.

A population health team might use those signals to prioritize:

• Earlier follow-up
• Medication management
• Care coordination
• Disease-specific education
• Remote monitoring
• Specialist referral
• Social support services

The appropriate intervention depends on the patient’s circumstances.

The predictive model provides a signal; the care team determines the response.

That distinction helps prevent analytics from becoming a substitute for clinical reasoning.

Model Accuracy Is Not the Only Measure of Success

Healthcare organizations sometimes evaluate predictive models primarily through technical measures such as sensitivity, specificity, discrimination, or predictive value.

Those metrics are important.

But healthcare leaders should also ask whether the model improves real-world outcomes.

A highly accurate model has limited operational value if clinicians cannot access its results or if care teams do not have the resources to act on identified risks.

Dr. Papin’s operational perspective suggests evaluating predictive analytics through a broader framework:

Prediction → Identification → Intervention → Follow-up → Outcome

If the process stops after prediction, the organization has created another reporting mechanism rather than a preventive-care system.

Governance and Bias Cannot Be Ignored

Predictive analytics also introduces important governance considerations.

Healthcare algorithms can reflect limitations or biases present in the data used to develop them. Differences in data availability, healthcare access, documentation practices, and historical utilization can influence model performance across patient populations.

The National Academy of Medicine and other healthcare organizations have emphasized the importance of responsible data use, transparency, evaluation, and appropriate oversight when applying artificial intelligence and advanced analytics in healthcare.

Organizations therefore need governance processes that address:

• Data quality
• Model validation
• Performance monitoring
• Patient privacy
• Security
• Bias and health equity
• Clinical oversight
• Appropriate human review

Predictive analytics should support clinical decision-making without creating the assumption that an algorithm is always correct.

Building Analytics Into the Workflow

The final challenge is operational integration.

A predictive model that exists only inside an analytics department is unlikely to change patient outcomes.

Its insights need to reach the people responsible for patient care.

That may require integration with electronic health records, care-management platforms, physician dashboards, population health systems, or other operational tools.

Even then, organizations need clearly defined workflows.

If a patient is identified as high risk, who receives the alert?

Who contacts the patient?

How quickly should the intervention occur?

What information does the clinician need?

How is the intervention documented?

How is the outcome measured?

These questions turn analytics from a technology project into an operational capability.

The Strategic Opportunity for Value-Based Care

Predictive analytics has particular relevance as healthcare moves toward greater financial accountability.

Under fee-for-service models, organizations may have limited financial incentives to prevent every future utilization event.

Under accountable care and other risk-based arrangements, the economic importance of prevention becomes more apparent.

Organizations that can identify rising-risk patients, close care gaps, coordinate services, and prevent avoidable deterioration may be better positioned to manage both quality and total cost of care.

For Dr. Papin, predictive analytics should therefore be viewed as part of the infrastructure supporting value-based healthcare — not as a standalone technology investment.

Looking Ahead

Healthcare has become increasingly sophisticated at measuring what happened.

The next challenge is becoming better at anticipating what may happen and responding before the situation becomes more difficult to manage.

For Joseph Papin, MD, predictive analytics reaches its full potential when it moves beyond dashboards and retrospective reporting into everyday preventive-care operations.

The strongest healthcare organizations will not necessarily be those with the largest datasets or the most complex algorithms.

They will be the organizations capable of translating reliable predictions into timely clinical action.

Prediction is the starting point.

Prevention is the objective.