AI and Stop Loss Underwriting: What We’re Getting Right and What We May Be Missing

A recurring question across the healthcare risk landscape is whether AI will meaningfully change the role of the stop loss underwriter. A growing view suggests that while AI will improve operational efficiency, underwriting itself remains largely a human activity, perhaps augmented by AI.

That perspective highlights the realities underwriters face. Claims data is often fragmented and inconsistent. High-cost claims sometimes stem from rare, complex clinical events, other times from entirely predictable ones. Pricing requires judgment, negotiation, nuance and accountability. Each employer group also brings a unique mix of demographics, geography and plan design that largely resists standardization.

These valid observations reinforce an important point: underwriting is not simply a data exercise. It is grounded in experience, context and informed judgment.

At the same time, this view reflects how both AI and healthcare risk are currently understood. At Evidium, we see a broader shift taking place. As healthcare risk becomes more clinically driven, the ability to connect medical knowledge with financial outcomes is becoming increasingly important to how risk is evaluated.

Where AI Fits in Underwriting Today

Today, AI is primarily used to improve efficiency within underwriting workflows. It helps with the RFP process, cleans and normalizes data, identifies high-cost claimants, generates baseline pricing assumptions and helps streamline renewal analysis.

These applications are valuable and help reduce manual effort, allowing underwriters to focus more on actual decision-making. In a market where margin discipline and stricter underwriting standards are increasingly non-negotiable, this kind of investment in better underwriting tools is becoming a competitive necessity.

However, these tools and approaches do not fundamentally change how risk is actually understood. Most approaches still rely on historical claims data and pattern recognition. They are retrospective, focusing on what has already happened and projecting it forward.

This is where limitations begin to emerge.

A Shift in the Nature of Healthcare Risk

Healthcare risk is increasingly concentrated and clinically driven. A small number of conditions and treatment pathways account for a disproportionate share of costs. These costs are not random. They are shaped by geography, income, disease progression, treatment decisions, emerging therapies and other patient-specific factors.

For example, two patients with the same diagnosis may follow very different care paths. Many conditions progress in stages that introduce escalating costs over time. Newer therapies, including biologics and cell and gene treatments, follow defined clinical protocols with significant financial impact.

These cases are often described as outliers. In reality, they are clinically explainable and predictable, even if they are not fully visible within traditional claims-based models.

The Gap in Risk Visibility

The industry is right to emphasize the importance of human judgment. However, judgment depends in large part on visibility. Claims data shows what has already happened. It reflects utilization and cost, but not why those events occurred in the first place or how they are likely to evolve over time.

Healthcare costs begin in the clinic, driven by disease progression, treatment pathways and clinical decisions. Without a structured way to understand these drivers, underwriting decisions are often made with limited visibility into the true nature of risk.

Evidium Expanding the Role of Healthcare Intelligence

This is where a different approach to healthcare intelligence becomes important.

Evidium is building the computational medical intelligence layer for healthcare — models of what causes what. Rather than relying only on historical claims data, Evidium makes medical knowledge computational: projecting patients across disease states, transitions, and intervention pathways that are actionable, traceable, and grounded in clinical evidence. Crucially, this also means that patients are no longer invisible to the risk equation — their clinical reality, and the trajectory of their care, becomes legible.

This allows underwriters and risk managers to move beyond identifying high-cost claims after they occur, and more toward understanding the clinical drivers behind those claims before costs fully materialize.

As a result, stakeholders can understand not only what has happened, but why it is happening and how it is likely to develop. This connects clinical decisions directly to financial outcomes and improves both risk assessment and long-term cost visibility. Importantly, this insight remains transparent and explainable and further allows one to better protect for it.

The Evolution of Underwriting and Healthcare Risk

Viewed through this lens, the role of the underwriter does not diminish. It evolves alongside a broader shift in how healthcare risk is managed. AI will continue to improve efficiency, but its greater value lies in strengthening human learning, decision-making and nuanced judgment. For organizations focused on margin over growth, that means investing in tools that improve the quality of every underwriting decision, not just the speed.

As risk becomes more complex and clinically driven, traditional approaches based on historical averages are being supplemented by deeper clinical understanding.

Underwriters will be better equipped to understand the drivers behind high-cost risks, anticipate how conditions may progress, evaluate treatment pathways and make more informed decisions.

Organizations that integrate clinical insight into their risk management frameworks will be better positioned to identify risk (and its severity) earlier, price it more accurately, improve capital efficiency and manage volatility more effectively.

In this model, AI does not replace judgment. It further enhances it.

The Path Forward

The view that AI will not replace stop loss underwriters is both reasonable and widely shared. The greater opportunity however, lies in how AI can further improve the quality of decision-making within the role.

While it will require the future underwriter to think and operate differently, the future of underwriting itself is not less human involvement but better-informed judgment. Those who treat clinical intelligence as a strategic investment rather than a cost will be better equipped to underwrite with discipline and precision that both current and future markets demand.  As healthcare continues to evolve, and that pace of change accelerates, those who combine experience and expertise with computational medical intelligence will be best positioned to navigate what comes next — and in doing so, make decisions that are better for the system, and better for patients.

Contact us today to learn how Evidium's computational medical intelligence can help you understand the clinical drivers behind risk — and make decisions that are better informed, more precise, and grounded in what's actually happening for patients.

About the Author

Rajiv Sood, GM Insurance & Risk, Evidium, is a distinguished and forward-thinking C-Suite leader, strategist and operational executive. He brings nearly forty years of extensive strategic planning, transformational leadership, business development and growth experience helping organizations breakthrough and elevate the enterprise to another level. He is recognized as a highly sought-after public speaker, global healthcare, accident, health and medical risk, insurance, reinsurance and insure tech expert, and global service provider operations leader. Rajiv has been honored to help organizations drive business value, propel the strategic vision and accelerate growth transforming businesses into thriving enterprises. He demonstrates a proven track record of success championing continuous improvement, unlocking the potential to accelerate growth and delivering on ambitious goals.