Why Healthcare’s AI Future Depends on Measuring What Matters to Patients

Healthcare systems worldwide are facing a paradox. Despite unprecedented advances in biotechnology, digital health, precision medicine, and artificial intelligence, healthcare costs continue to rise while outcomes remain highly variable between providers, regions, and patient populations. Global healthcare spending is projected to approach $10 trillion annually, yet policymakers and healthcare leaders still struggle to answer a deceptively simple question: how do we know whether patients are actually getting better?

For decades, most healthcare systems measured success through activity. Hospitals tracked admissions, procedures, and reimbursement claims. Governments monitored utilization rates. Insurers focused on costs and efficiency. While these metrics remain important for managing healthcare systems, they reveal surprisingly little about whether patients experience less pain, regain mobility, return to work, or enjoy a better quality of life following treatment.

As healthcare systems increasingly transition toward value-based care models, attention is shifting toward a different metric: patient outcomes. According to Jennifer Bright, President and CEO of the International Consortium for Health Outcomes Measurement (ICHOM), this evolution may prove critical not only for healthcare quality, but also for the future development of healthcare artificial intelligence.

“Essential to smart and high-quality healthcare delivery is a relentless focus on outcomes,” Bright says. “Specifically, the patient outcomes that matter most to the patients themselves.”

For more than a decade, ICHOM has worked with healthcare organizations worldwide to develop standardized outcome measurement frameworks spanning 47 health conditions and multiple stages of life. The organization’s mission reflects a broader industry realization that measuring healthcare activity is no longer enough. To improve healthcare, systems must understand the results patients actually experience.

The True Benchmark of Quality: Elevating Patient-Reported Outcomes

Modern healthcare generates an extraordinary volume of data. Electronic medical records, imaging systems, laboratory tests, wearable devices, and reimbursement claims collectively produce billions of data points every day. Yet many healthcare leaders argue that one of the industry’s most valuable data sources remains underutilized: the patient’s own voice.

This challenge has become increasingly visible as healthcare systems attempt to move from fee-for-service reimbursement toward value-based care. While hospitals can easily track how many procedures were performed or how long a patient remained hospitalized, determining whether treatment meaningfully improved a person’s life is considerably more difficult.

To bridge that gap, healthcare organizations around the world are increasingly adopting Patient-Reported Outcome Measures (PROMs). These standardized assessments allow patients to directly report their pain levels, physical function, mental wellbeing, and overall quality of life before and after treatment.

By capturing outcomes from the patient’s perspective, PROMs provide insights that clinical measurements alone often miss. Two patients may receive identical treatments and achieve similar laboratory results, yet experience dramatically different recoveries and quality-of-life outcomes.

Despite growing recognition of their value, implementing PROMs remains a significant challenge worldwide.

The Clinician Burden Crisis

Healthcare professionals are already facing unprecedented levels of administrative pressure. Across many healthcare systems, physicians and nurses spend a substantial portion of their day completing documentation, compliance requirements, and reporting tasks. Introducing outcome measurement programs can initially be perceived as another administrative burden rather than a clinical improvement tool.

As a result, healthcare organizations frequently encounter resistance when attempting to integrate PROM collection into routine practice.

The Data Sovereignty Challenge

Healthcare data remains fragmented across institutions, regions, and jurisdictions. Privacy regulations, technical limitations, and organizational silos often prevent outcome information from being shared or analyzed effectively. Valuable patient insights may remain locked within individual systems, limiting opportunities for broader learning and quality improvement.

Yet Bright argues that organizations frequently experience a dramatic shift once patient-reported outcomes become embedded within clinical workflows.

Rather than relying solely on assumptions about treatment effectiveness, clinicians gain direct visibility into how patients are recovering and where care pathways can be improved. Outcome measurement becomes more than a reporting exercise—it becomes a tool for better decision-making.

“When care teams begin to see patient-reported outcomes in real time, it changes how they practice,” Bright explains. “They can see the impact of their work more clearly and identify opportunities to improve care.”

The Digital Paradox: Why AI Needs a Human Lens

Few technologies have generated as much excitement in healthcare as artificial intelligence. Investors have poured billions of dollars into AI-driven diagnostics, predictive analytics, clinical decision support systems, and operational optimization platforms. Healthcare organizations increasingly view AI as a potential solution to workforce shortages, rising costs, and growing patient demand.

Yet Bright believes the industry’s enthusiasm has obscured a fundamental question: what exactly are these systems learning?

Many healthcare AI models are trained primarily on claims data, procedural records, billing codes, and historical utilization patterns because these datasets are abundant, standardized, and relatively easy to access. However, these data sources were never designed to measure patient outcomes. They were designed to support reimbursement and administrative functions.

As a result, AI may become highly effective at predicting healthcare activity without necessarily understanding healthcare value.

“AI today is entirely about how these models are trained,” Bright warns. “If they are trained solely on procedural or claims data, we run the serious risk of hard-coding systemic biases directly into the technology.”

This concern extends far beyond technical performance. If certain populations historically experienced barriers to healthcare access, delayed diagnoses, or reduced utilization of specific services, AI systems trained on those historical patterns may inadvertently reproduce those inequities in future recommendations.

To avoid automating the shortcomings of existing healthcare systems, Bright argues that AI development must incorporate outcome-focused data and patient perspectives.

Traditional AI vs. Value-Based AI

The distinction may become increasingly important as healthcare organizations rely on AI to guide clinical decisions, resource allocation, and reimbursement strategies.

Why Human Oversight Remains Essential

The growing sophistication of healthcare AI has sparked ongoing debates about autonomy, accountability, and decision-making. As algorithms become more capable, questions naturally emerge regarding how much responsibility should be delegated to machines.

Bright does not see AI as a replacement for clinicians. Instead, she views it as a powerful tool that must remain grounded in human expertise and lived patient experience.

To achieve that balance, healthcare organizations need robust human-in-the-loop governance structures capable of continuously evaluating algorithmic performance.

Effective oversight requires input from two distinct perspectives:

• The Clinical Expert: Healthcare professionals provide the medical expertise necessary to interpret AI outputs within complex clinical environments. They understand disease biology, patient variability, treatment limitations, and contextual factors that algorithms may not fully capture. Clinical oversight helps ensure that AI recommendations remain medically appropriate and aligned with established standards of care.
• The Lived-Experience Expert: Patients contribute an equally important perspective. Individuals who have personally experienced a disease or condition can evaluate whether healthcare technologies are producing outcomes that align with real-world needs, expectations, and quality-of-life priorities. This perspective helps ensure that healthcare innovation remains centered on the people it is intended to serve.

Together, these viewpoints create a governance framework that balances technological advancement with human judgment, helping prevent AI systems from drifting away from patient-centered care.

Blueprint for Execution: Five Strategies for Value-Based Healthcare

While the principles of value-based healthcare are increasingly accepted, implementation remains uneven across global healthcare systems. Many organizations understand the need to improve outcome measurement but struggle to operationalize meaningful change.

Drawing from ICHOM’s experience supporting healthcare organizations around the world, Bright highlights five practical strategies for accelerating the transition toward outcome-driven care.

1. Embed PROMs Directly Into Clinical Workflows: Patient-reported outcomes should be integrated into routine care rather than treated as standalone administrative exercises. Outcome collection becomes most valuable when clinicians can immediately apply the information to treatment decisions.
2. Build Federated Learning Networks: Decentralized AI frameworks allow institutions to collaborate and generate insights without transferring sensitive patient data. Federated learning can help overcome privacy barriers while supporting large-scale innovation.
3. Establish Human-in-the-Loop AI Governance: Every healthcare algorithm should undergo ongoing evaluation by both medical professionals and patients. Continuous oversight helps ensure fairness, relevance, and clinical validity.
4. Treat Patient Data as a Shared Asset: Healthcare organizations should provide patients with transparent access to outcome information and recovery metrics, transforming them from passive recipients of care into active participants in healthcare decision-making.
5. Standardize Outcome Measurement: Globally recognized outcome frameworks allow healthcare organizations to benchmark performance, compare results, and share best practices across borders. Standardization creates the foundation necessary for meaningful learning at scale.

Turning Outcomes Into Action

Healthcare leaders frequently describe data as the fuel of innovation. Yet Jennifer Bright argues that the future of healthcare will depend less on the amount of data collected and more on whether organizations collect data that reflects what patients truly value.

As AI becomes increasingly embedded in healthcare delivery, reimbursement models continue evolving, and health systems search for sustainable ways to improve outcomes, patient-reported measures may emerge as one of the most strategically important datasets in medicine.

“We are all at the front of a very fast train,” Bright reflects. “International collaboration and shared learning are vital.”

Ultimately, the future of healthcare may depend on whether organizations can move beyond measuring what is easiest to capture and instead focus on what matters most. For decades, healthcare has excelled at measuring activity. The next era may be defined by measuring outcomes—and ensuring that both humans and machines learn from them.

Because in the end, healthcare quality is not defined by the number of services delivered. It is defined by whether patients’ lives improve.

Jennifer Bright (featured), President and CEO of ICHOM, shares her perspective on the future of healthcare quality, arguing that patient outcomes—not procedures, claims, or utilization rates—should become the foundation for both value-based care and healthcare AI. Image: GeneOnline

Author

Bernice Lottering