Despite major advances in diagnostic technologies and therapeutics, preventable vision loss from glaucoma remains common. In many cases, the disease is detected late, progression is recognized too slowly, or treatment escalation occurs only after functional decline has already begun.
This is not due to a lack of information. Glaucoma care is increasingly rich in an overwhelming volume of data. The challenge lies in translating that information into timely, individualized decisions that preserve meaningful vision over a patient’s lifetime.
Why Glaucoma Care Still Falls Behind
Delays occur at multiple levels, starting with the disease itself. Early glaucoma is typically asymptomatic, and many patients do not notice vision changes until the disease is advanced when opportunities to preserve function may already be reduced.
Complicating matters, glaucoma has no single pathognomonic sign or symptom. Diagnosis often depends on a constellation of clinical findings, structural and functional diagnostic testing, and critically longitudinal data to confirm both disease presence and progression. Subtle abnormalities may be missed, particularly when clinicians rely too heavily on a single modality or interpret results without trend context. Additionally, there remains limited consensus on definitions of early glaucoma and on structural and functional progression endpoints.
Even after diagnosis, uncertainty persists. Progression is unpredictable, treatment response varies, and adherence and follow-up reliability differ widely. Escalation decisions about whether to intensify medications, pursue laser, or move toward surgery are often made with incomplete information and limited ability to forecast risk and pace of decline
Disconnected Data
Today’s glaucoma practice generates continuous inputs: intraocular pressure (IOP) trends across visits, optic nerve assessments, structural metrics on optical coherence tomography (OCT), visual field testing, medication history and tolerability, laser procedures, surgical outcomes, and adherence patterns.
At the same time, patients are increasingly complex. Aging, multimorbidity, polypharmacy, systemic disease, and social determinants, transportation, cost, caregiver support, health literacy directly influence follow-up and treatment success.
Yet these data are often fragmented across imaging platforms, perimetry devices, and the EHR. Even when all the information is technically available, it may not be accessible in a unified, decision-ready format. This fragmentation creates blind spots: difficulty integrating structure and function, limited ability to quantify progression risk, and uncertainty about when escalation is truly warranted.
Why Glaucoma Is a “Decision Problem”
Glaucoma care is not a single diagnostic moment or a one-time treatment choice. It is a long-term process shaped by high-stakes decisions repeated over years. During each visit, clinicians must decide:
- Is this glaucoma or physiologic cupping?
- Is change real, or simply a result of test variability?
- If progression exists, how fast?
- What target IOP is appropriate now and does it need to change?
- When should therapy be initiated or escalated?
- When should laser or surgery be considered?
- How frequently should monitoring and testing occur?
- Who is safe to follow conservatively, and who is at imminent risk of functional loss?
These questions require synthesis of structural and functional findings, clinical context, and longitudinal trends often under uncertainty, especially early in disease when signals may be subtle or discordant. The stakes are high: progression is irreversible, and patients may remain asymptomatic until substantial damage has occurred.
The Personalization Gap
Despite the complexity of glaucoma decision-making, care remains largely disease centered, built around diagnostic categories and standardized guidelines designed for consistency and efficiency. This approach reduces variation and supports baseline standards, but it can implicitly assume that patients with the same diagnosis share similar risks and goals.
In contrast, patient-centered care emphasizes individualized risk assessment, shared decision-making, and alignment of management strategies with each patient’s circumstances and functional goals. In glaucoma, optimal care requires both perspectives: understanding disease mechanisms while accounting for patient-specific factors such as age, genetics, systemic disease, systemic medications, comorbidities, lifestyle, and social and behavioral determinants that influence adherence and outcomes.
The challenge is scale. Personalization is difficult to deliver consistently across busy clinics. This is where decision intelligence becomes clinically relevant.
Clinical Applications of Decision Intelligence
Decision intelligence is a multi- disciplinary approach that transforms complex data into practical clinical decisions by combining data analytics, artificial intelligence, and decision science. In health care, it refers to systematic data-driven insights that support rather than replace clinical judgment, particularly when decisions are high-stakes and made under uncertainty.1
Glaucoma is a prime use case. Diagnosis and management rarely depend on a single test or visit. Clinicians must synthesize IOP trends, optic nerve findings, OCT structural metrics, visual field performance, treatment history, and longitudinal change while accounting for variability and patient-level risk.
Decision intelligence bridges the gap between data, decisions, and outcomes by integrating multimodal, longitudinal information into a unified, analytics-driven approach. Importantly, it can incorporate data beyond traditional glaucoma testing, including systemic disease and medication profiles, comorbidity and laboratory data, EHR documentation, adherence history, patient-reported outcomes, and at a population level claims/utilization data, social determinants of health, genetic risk markers, and real-world evidence from remote monitoring. The goal is a more complete view of risk and progression to support proactive, personalized care.
Enhancing Diagnostic Accuracy
Early detection remains challenging because structural and functional signals may be subtle, borderline, or discordant. Decision intelligence improves diagnostic confidence by integrating OCT, perimetry, and clinical history into a unified framework while also incorporating systemic disease, systemic medications, genetics, and environmental risk factors.
In practice, this may look like an OCT trend showing consistent sectoral retinal nerve fiber layer or ganglion cell complex thinning that aligns with an emerging cluster of depressed points on visual field testing supporting earlier recognition of true progression before global indices clearly worsen. Decision intelligence tools can help clinicians visualize these multimodal trends faster and more consistently, improving confidence in early disease.
Personalized Patient Management and Treatment
Progression risk and treatment response vary widely. Decision intelligence can leverage longitudinal data on serial visual fields, OCT, and IOP trends to identify patterns linked to rapid progression, inadequate pressure control, intolerance, or higher likelihood of requiring laser or surgery. These models may also help predict response to specific medical and surgical treatments and guide selection of the next best step (additional medications, laser, MIGS, filtration surgery), reducing trial-and-error escalation. Predictive modeling work highlights the potential to anticipate functional decline and align treatment intensity with projected risk.2-3
Optimizing Monitoring and Follow-Up
Follow-up interval and testing frequency decisions remain difficult. Uniform schedules can overtest stable patients while undermonitoring those at highest risk. Decision intelligence supports risk-based monitoring recommendations aligned with severity, progression patterns, and patient-specific risk factors improving patient experience, clinic efficiency, and safety in the setting of constrained capacity.
Supporting Integrated, Team-Based Care
Glaucoma care is often fragmented across platforms and workflows. Decision intelligence can integrate these data streams into a cohesive patient view, supporting coordinated team-based care, particularly proactive identification of patients at risk of progression, delayed follow-up, or loss to follow-up and prioritizing outreach and escalation decisions.
Evaluating Treatment Outcomes and Real-World Effectiveness
Decision intelligence supports continuous improvement by analyzing outcomes across large populations to understand which interventions work best for which patients and under what circumstances. Real-world care includes variable adherence, comorbidities, mixed disease severity, and inconsistent follow-up. By evaluating how baseline characteristics, adherence, comorbidities, and demographics influence outcomes, practices can refine clinical pathways, reduce variation, and optimize long-term results. These tools can also enable learning-system feedback loops by tracking outcomes over time and comparing performance across procedures, surgeons, and care settings, supporting quality improvement and more tailored counseling.
Optimizing Resource Allocation and Population Management
As glaucoma prevalence rises, clinics face increasing pressure on testing capacity, follow-up availability, and surgical scheduling. Decision intelligence can help allocate resources more effectively by identifying patients needing urgent evaluation or intensified monitoring and those who can safely extend follow-up intervals.
It can also support value-based decision-making by incorporating cost-effectiveness and quality-of-life considerations into treatment planning. For example, Sood et al demonstrated how modeling approaches can guide resource utilization through cost-effectiveness analyses of minimally invasive trabecular meshwork stents combined with phacoemulsification, as well as prophylactic laser peripheral iridotomy in primary angle-closure suspects, integrating both economic and quality-of-life outcomes to inform complex clinical decisions.4-5 At the population level, risk-based dashboards and registries can improve access, reduce missed progression, and align resources with patients who need them most.
Challenges and Limitations
Several barriers must be addressed before decision intelligence can be implemented reliably at scale. Data quality is fundamental. Glaucoma testing is variable and affected by artifacts, patient performance, device differences, and inconsistent documentation. Without safeguards, models trained on noisy inputs may produce misleading outputs.
A related challenge is the lack of universal consensus on glaucoma definitions and endpoints: particularly, what constitutes clinically meaningful change in structural and functional measures. Disagreement persists on how best to define progression, reconcile discordant OCT and visual field signals, and select endpoints across clinical care, research, and regulatory settings. Without standardized definitions and validated endpoints, it is harder to train, compare, and deploy tools consistently and transparently.
Interoperability remains a major obstacle across OCT platforms, perimetry systems, and electronic health records. Bias and generalizability also matter; models trained on nonrepresentative data sets may perform poorly in diverse populations and reinforce disparities.6 Finally, adoption depends on clinician trust and workflow fit. Tools must be explainable, transparent, and easy to use, with governance to address privacy, ethics, and accountability. Decision intelligence should strengthen clinical reasoning, not override it.
Future Directions
The next phase of glaucoma care will likely involve broader integration of decision intelligence across screening, diagnosis, monitoring, and treatment planning. Predictive modeling may improve forecasts of progression risk and functional decline, enabling earlier escalation for patients most likely to lose vision.
Remote monitoring and teleophthalmology may expand access and reduce delays, especially for patients facing geographic, socioeconomic, or mobility barriers. Incorporating patient-reported outcomes, social determinants of health, and precision medicine principles may further strengthen personalization by aligning strategies with adherence, functional needs, and quality-of-life priorities. As decision intelligence systems learn from real-world data, glaucoma care has the potential to evolve into a learning health system one that improves continuously over time.
Conclusion
Glaucoma management is data rich, but its most important decisions, diagnosis, monitoring intensity, escalation, and surgical timing depend on synthesizing complex information over time. Decision intelligence offers a practical path forward: integrating multimodal and longitudinal data, incorporating systemic and social risk factors, and translating trends into actionable insights to support faster, more consistent, and more personalized care.
The goal is not to replace clinicians with algorithms. The goal is to reduce uncertainty, improve consistency, and ensure decisions are made early enough to preserve meaningful vision across a lifetime and prevent avoidable blindness. GP
References
1. Dey S. Advancing Healthcare Through Decision Intelligence: Machine Learning, Robotics, and Analytics in Biomedical Informatics. 1st ed. Elsevier; 2025.
2. Liu J, Sood S, Razavian N, et al. Predicting progression of glaucoma based on visual field tests. Invest Ophthalmol Vis Sci. 2023;64(8):335.
3. Si W, Lin V, Sun B, et al. Reliable and interpretable visual field progression prediction with diffusion models and conformal risk control. In: Gee JC, Alexander DC, Hong J, et al, eds. Medical Image Computing and Computer-Assisted Intervention—MICCAI 2025. Vol 15974. Lecture Notes in Computer Science. Springer Nature; 2026:543-552. doi:10.1007/978-3-032-05182-0_53
4. Sood S, Heilenbach N, Sanchez V, Glied S, Chen S, Al-Aswad LA. Cost-effectiveness analysis of minimally invasive trabecular meshwork stents with phacoemulsification. Ophthalmol Glaucoma. 2022;5(3):284-296. doi:10.1016/j.ogla.2021.09.006
5. Sood S, Sanchez V, Heilenbach N, Al-Aswad LA. Cost-effectiveness of prophylactic laser peripheral iridotomy in primary angle-closure suspects. Ophthalmol Glaucoma. 2023;6(4):332-341. doi:10.1016/j.ogla.2022.10.005
6. Ong J, Jang KJ, Baek SJ, et al. Development of oculomics artificial intelligence for cardiovascular risk factors: a case study in fundus oculomics for HbA1c assessment and clinically relevant considerations for clinicians. Asia Pac J Ophthalmol (Phila). 2024;13(4):100095. doi:10.1016/j.apjo.2024.100095.