Traditional PROMs questionnaires, widely adopted in eye care clinics and clinical trials, are typically criticized for their length, repetitiveness, and lack of adaptability. These limitations hinder effective real-time clinical decision-making and reduce patient engagement, ultimately compromising the utility of the data collected. Recognising this challenge, PROMinsight has engineered an AI-powered platform that harnesses sophisticated psychometric algorithms, along with extensive ophthalmic datasets gathered from thousands of patients, to tailor question delivery in a patient-specific manner. This personalised interrogation not only shortens completion times but also enhances the relevance and accuracy of the information obtained.

At the core of PROMinsight’s technology is its computerised adaptive testing capability, which dynamically adjusts each subsequent question based on the individual’s previous responses. This real-time adaptation enables the system to hone in on assessing the specific dimensions of a patient’s visual health and quality of life that are most pertinent, thereby bypassing irrelevant or redundant queries. The resultant data has greater clinical value and enables health professionals to intervene with a nuanced understanding of how ocular diseases and treatments affect daily functions such as mobility, emotional well-being, and patient comfort.

The development of PROMinsight’s capabilities was spearheaded by Professor Ecosse Lamoureux, who integrates expertise in population health and ophthalmic research at both Duke-NUS Medical School and the Singapore Eye Research Institute. His team has meticulously developed CATs tailored to a spectrum of eye conditions prevalent in Asian populations, including diabetic retinopathy, macular edema, cataracts, myopia, glaucoma, and age-related macular degeneration. By leveraging a robust framework of established psychometric principles and intensive empirical data, these tests deliver quantifiable metrics that reflect patient experiences with unprecedented clinical precision.

Professor Lamoureux emphasizes that AI-enhanced PROMs transcend the boundaries of conventional questionnaires by actively listening to patients’ lived experiences and adapting the measurement process accordingly. The CAT methodology equips ophthalmologists with a versatile and powerful tool that captures the nuanced impact of vision impairment beyond visual acuity alone. Moreover, the ability to integrate these tools seamlessly into clinical workflows—supported by cloud-based administration and instantaneous scoring—facilitates real-time clinical decisions informed by comprehensive patient-reported data.

Among the proprietary CAT-based instruments licensed to PROMinsight are MyoRI-CAT for evaluating interventions in myopia, GlauCAT-Asian tailored to glaucoma management in Asian clinical contexts, and MacCAT, designed for patients with age-related macular degeneration. These domain-specific tools embody advanced item response theory models, thereby enabling clinicians to extract meaningful quality of life (QoL) insights that correlate with both patient symptoms and treatment responsiveness.

PROMinsight’s innovation underwent rigorous validation during a year-long implementation study conducted at the Singapore National Eye Centre, which focused on patients receiving treatments for diabetic macular edema, glaucoma, and cataract. The results were striking: patients demonstrated over 80% engagement with the CAT system and reported highly positive user experiences, noting the questionnaires addressed dimensions of daily life often overlooked in routine care. The success of this pilot is propelling plans for broader deployment across multiple departments within the institution.

Feedback from clinicians underscores the clinical value of integrating PROMinsight’s CATs into practice. For instance, quality of life scores captured by these assessments revealed discrepancies between objective visual measures and patient well-being, illuminating hidden deficits that would otherwise go undetected. Such insights have the potential to refine treatment regimens and patient counseling by foregrounding personalized patient experiences alongside traditional clinical indicators.

Beyond Singapore, PROMinsight is forging international collaborations with ophthalmologists, academic medical centers, pharmaceutical companies, and regulatory bodies including the FDA and EMA. These partnerships underscore the platform’s scalability and applicability for clinical trials of novel ophthalmic therapies as well as routine care. Notably, a recent substantial non-dilutive investment from a leading ophthalmic pharmaceutical firm further validates the commercial and clinical impact potential of PROMinsight’s technologies.

Dr. Eva Fenwick, CEO and Co-Founder of PROMinsight and an adjunct professor at Duke-NUS, articulates the platform’s dual promise: enabling rigorous collection of patient-reported data to accelerate pharmaceutical research, while simultaneously equipping eye clinics worldwide with an advanced toolset for monitoring and improving patient outcomes. This convergence of research and clinical utility marks a significant advance in managing and treating eye conditions with patient-centered precision.

The broader significance of PROMinsight’s development lies in its contribution to the evolving paradigm of value-based care—where treatment success is measured not only by clinical parameters but also by the outcomes that matter most to patients. In an era increasingly focused on healthcare results and patient engagement, intelligent tools like PROMinsight’s CATs are becoming indispensable for capturing the multifaceted reality of patient health, fostering more informed clinical decisions and targeted therapeutic strategies.

As AI and data analytics continue to permeate the healthcare landscape, PROMinsight exemplifies how domain-specific adaptation of these technologies can yield transformative improvements. By turning subjective patient experiences into robust, actionable data, this approach exemplifies the future of personalized medicine in ophthalmology, offering hope for improved patient satisfaction and clinical outcomes on a global scale.

Subject of Research:
People

Article Title:
AI-Driven Adaptive Testing Revolutionizes Patient-Reported Outcomes in Vision Care

News Publication Date:
7 April 2026

Web References:
www.duke-nus.edu.sg
www.seri.com.sg

Image Credits:
PROMinsight

The research team’s ingenuity stems from their ability to harmonize two powerful domains: data-driven machine learning algorithms and centrifugal microfluidics—a miniaturized technology that enables rapid processing of biological samples. Centrifugal microfluidics uses controlled spinning forces to precisely manipulate small volumes of fluids, dramatically shortening assay times without compromising accuracy. By harnessing this technology, the researchers developed a compact, portable platform capable of analyzing complex biological markers associated with the systemic inflammatory response characterizing sepsis. What distinguishes this platform from existing methodologies is its seamless integration with machine learning models trained on vast datasets of clinical variables, granting it unique predictive accuracy that outperforms current gold standards.

The intrinsic challenge of sepsis lies in its heterogeneity; it manifests through a complex interplay of host immune responses and pathogenic factors that fluctuate dynamically. Conventional laboratory diagnostics often involve lengthy processing times, and clinical judgment alone can lead to delayed or missed diagnoses. The newly developed platform addresses these shortcomings by providing a rapid point-of-care solution that delivers robust predictions within minutes. Blood samples obtained at the patient’s bedside are processed through the centrifugal device, extracting critical biochemical signatures that feed into the AI algorithm. This system not only identifies early signs of sepsis but also stratifies patients according to risk, thereby informing more personalized and timely therapeutic interventions.

One key technical feature of the platform is its sophisticated machine learning architecture, which includes ensemble methods to improve predictive stability and generalizability across diverse patient populations. The researchers utilized comprehensive training sets derived from multi-center clinical data, incorporating variables such as cytokine levels, vital signs, and patient demographics. This approach ensures that the algorithm adapts to the nuanced presentations of sepsis seen across different healthcare settings and patient profiles. Furthermore, rigorous cross-validation protocols were employed to fine-tune the model’s sensitivity and specificity, pushing the boundaries of diagnostic confidence and minimizing false positives and negatives.

Equally impressive is the engineering feat underlying the centrifugal microfluidics device itself. The platform employs a bespoke disc design that channels the biological sample through multiple reaction chambers as it spins, enabling simultaneous multiplexed assays. This centrifugal force-driven fluid transport negates the need for bulky pumps or valves, significantly reducing device complexity and size. Within these microchambers, reagents react swiftly with blood analytes to generate quantifiable signals that are electrochemically or optically detected. The miniaturization and automation inherent in this design substantially reduce operator demands and variability, paving the way for widespread clinical adoption in resource-limited and emergency settings alike.

The integration of these two technologies culminates in a seamlessly automated workflow where sample preparation, reaction, signal detection, and data processing occur in tandem. The user interface was designed with clinicians in mind, featuring intuitive touchscreen controls and real-time data visualization that clearly convey sepsis risk levels. This immediacy is critical in acute care, where every minute counts. Real-world validation studies demonstrated that the platform consistently delivered predictions within 30 minutes of sample collection—an exponential improvement over traditional laboratory techniques that often take several hours. Such rapid turnaround empowers emergency physicians and intensivists to initiate early, targeted interventions that can be life-saving.

What makes the platform especially compelling is its scalability and adaptability. Because the microfluidic disc can be customized with different reagents, the system can potentially be expanded to detect other infectious or inflammatory conditions beyond sepsis, evolving into a versatile bedside diagnostic tool. Similarly, the AI algorithms are designed to continuously learn from new patient data, augmenting their predictive capabilities with ongoing clinical deployment. This dynamic feedback loop aligns with the vision of precision medicine, where diagnostics evolve in real time to accommodate emerging disease patterns and pathogen variants.

The potential global impact of this technology cannot be overstated. Sepsis is not confined by geography or socioeconomic boundaries, disproportionately affecting populations in low- and middle-income countries where rapid diagnostics are often unavailable. The portable nature of the platform, coupled with its minimal reliance on complex laboratory infrastructure, renders it ideally suited for deployment in under-resourced settings. By facilitating earlier detection and more accurate risk assessment, this device could drastically reduce sepsis-related morbidity and mortality worldwide, addressing a pressing unmet need in global health.

In addition to clinical advantages, the technology exemplifies the successful marriage between biomedical engineering and clinical informatics. The interdisciplinary collaboration between engineers, data scientists, and clinicians was paramount to navigating the complex path from concept to clinical proof-of-concept. The researchers emphasize that ongoing partnerships with healthcare providers will be essential to refining usability and ensuring regulatory compliance, which will ultimately govern widespread clinical adoption. Furthermore, extensive field trials are underway to evaluate impact on patient outcomes, cost-effectiveness, and integration into existing care pathways.

Beyond sepsis, this paradigm of coupling centrifugal microfluidics with machine learning heralds a new era for bedside diagnostics. As artificial intelligence and microengineering advance in tandem, we may witness a transformation in how acute diseases—including stroke, myocardial infarction, and infectious outbreaks—are detected and managed at the point of care. The platform serves as a template demonstrating that rapid, automated, and intelligent diagnostics can be accessible outside traditional laboratory settings, shifting diagnostic power directly into clinicians’ hands.

Ethical considerations surrounding the deployment of AI-driven diagnostic platforms also arise. Ensuring algorithmic transparency, guarding patient data privacy, and maintaining clinician oversight are critical factors addressed by the research team. The authors advocate for regulatory frameworks that balance innovation with safety, underscoring that machine learning supplements but does not replace clinical expertise. Transparency in algorithm development and continuous performance monitoring are vital to building trust among clinicians and patients alike.

Importantly, the technology exemplifies how microfluidic devices can be combined with artificial intelligence not just for predictive analytics but for enabling precision interventions. By rapidly identifying specific sepsis phenotypes and severity, the device could guide tailored antimicrobial therapy, fluid resuscitation strategies, and immunomodulatory treatments. This level of granularity in diagnostics promises to improve therapeutic efficacy while reducing the risk of overtreatment and antibiotic resistance—a persistent challenge in sepsis management.

Looking forward, the research team envisions integrating the platform with electronic health records and hospital information systems to establish seamless data flows and longitudinal patient monitoring. Such connectivity could facilitate continuous risk assessment, post-discharge surveillance, and real-time decision support across care transitions. The prospect of embedding AI-powered diagnostics within broader healthcare ecosystems signals an important step toward smarter, more responsive health systems.

To conclude, the innovative work by Malic, Zhang, Plant, and colleagues represents a milestone in confronting the global sepsis crisis. By harnessing the synergy between centrifugal microfluidics and advanced machine learning, they have created a powerful bedside diagnostic tool that promises to save countless lives through earlier detection and smarter intervention. As this technology moves from bench to bedside, it not only transforms sepsis care but also sets the stage for a new generation of intelligent medical devices with profound implications across healthcare.

Subject of Research: Bedside prediction and diagnosis of sepsis using a combined machine learning and centrifugal microfluidics platform.

Article Title: A machine learning and centrifugal microfluidics platform for bedside prediction of sepsis.

Article References:
Malic, L., Zhang, P.G.Y., Plant, P.J. et al. A machine learning and centrifugal microfluidics platform for bedside prediction of sepsis. Nat Commun 16, 4442 (2025). https://doi.org/10.1038/s41467-025-59227-x

Image Credits: AI Generated