Based in Cambridge UK, Qureight, a core imaging laboratory has been making significant progress since their inception in 2018. It has forged meaningful partnerships with pharma, biotech and contract research organisations. Their proprietary technology is accelerating lung and heart drug development using AI-powered imaging and data curation to optimize trial endpoints and patient stratification. The company also has a broader vision to explore rare and autoimmune diseases.
Damian Doherty, editor in chief of Inside Precsion Medicine, spoke with Simon Walsh Chief Scientific Officer at Qureight to explore the company’s powerful cutting-edge AI technology and its transformative potential. From lung and heart disease to broader applications, Walsh shares how Qureight aims to revolutionize clinical insights by building a more connected view of disease, redefining the future of precision medicine through AI-driven innovation.
Q: Qureight has established itself as a leader in applying deep learning to clinical data for lung and heart diseases. Could you elaborate on how your deep learning technology goes beyond traditional analytics to identify novel and more sensitive biomarkers from complex datasets, such as computed tomography (CT) scans and physiological data?
Walsh: Traditional imaging analytics rely on predefined metrics such as fibrosis extent or lung volume loss. However, fibrosis quantification alone does not capture the full prognostic signal that high-resolution computed tomography (HRCT) has to offer. Our approach uses deep learning models trained on human-annotated data to extract multi-compartment imaging biomarkers that reflect prognostically meaningful changes in the pulmonary vasculature, airway tree, and lung parenchyma. These biomarkers are reproducible, scalable, and trial-ready, and they reflect biological processes that standard measures like forced vital capacity (FVC) often miss. We can also incorporate measures of uncertainty, allowing us to flag areas of diagnostic ambiguity and enhance interpretability for sponsors. Beyond imaging alone, we integrate our biomarkers with other omics data, such as proteomics, to enhance risk stratification. Critically, we use these biomarkers to construct synthetic control arms in antifibrotic therapy trials, allowing more precise matching between cohorts, reducing placebo exposure, and accelerating signal detection—all while aligning with emerging regulatory frameworks.
Q: How does this accelerate the drug development process for your partners?
Walsh: Signal extraction from CT imaging is at the heart of how we optimize clinical trials. Our platform isolates and quantifies treatment effects within the specific anatomical compartments of the lung—parenchyma, airways, and vasculature—rather than relying solely on conventional metrics like FVC. This compartmental approach allows us to identify therapeutic signals that directly reflect a drug’s mechanism of action. For example, in a recent Phase II idiopathic pulmonary fibrosis (IPF) trial, our biomarkers revealed a larger treatment effect size associated with pulmonary vascular volume than FVC—precisely the compartment expected to respond based on the drug’s vascular mechanism. This level of mechanistic precision enables earlier go/no-go decisions, synthetic control arm design, and enhanced regulatory credibility. We also accelerate trials by enabling cohort enrichment: our biomarkers can identify patients with lightly progressive disease, who are most likely to demonstrate a therapeutic signal. By selecting the right patients and extracting the right signals from imaging, we help our partners reduce sample sizes, increase endpoint sensitivity, and compress timelines across the trial lifecycle.
Q: Your collaboration with Insilico Medicine on the IPF drug rentosertib has been a significant milestone. What specific insights did Qureight’s platform provide that supported the preliminary efficacy results and future trial expansion for a drug discovered using generative AI?
Walsh: Our collaboration with Insilico Medicine focused on validating the external generalizability of their Phase II trial results for rentosertib, a novel antifibrotic discovered using generative AI. The trial was conducted exclusively in a Chinese IPF population, raising important questions about how well the findings might translate to global regulatory and clinical settings. Using our global IPF datasets, we demonstrated that the Chinese cohort showed no material differences in key enrollment criteria compared to global IPF populations. This finding supported the argument that the treatment effect observed in the Chinese trial was not specific to a unique regional phenotype and could, in principle, generalize to the broader IPF population worldwide.
Q: What does this collaboration signify about the synergy between different AI applications in the drug development pipeline?
Walsh: The Insilico collaboration illustrates the power of linking upstream and downstream AI capabilities in drug development. Generative AI can identify novel therapeutic compounds like rentosertib, but success ultimately hinges on proving that these candidates work in real patients. That’s where Qureight comes in. Our role is to enable precision phenotyping and global applicability testing through advanced clinical and imaging biomarkers. In this case, Insilico discovered a novel IPF drug candidate; our platform validated that the trial population in China exhibited the same disease signature as IPF populations elsewhere in the world. This potentially derisks their expansion strategy and laid a scientific foundation for generalizing efficacy claims across diverse regulatory territories
Q: Beyond the use of deep learning for biomarker identification, Qureight has pioneered the use of “synthetic control arms.” Can you explain the concept behind this technology, its ethical and practical advantages, and how it is revolutionizing clinical trial design, particularly for rare and severe diseases like IPF?
Walsh: A synthetic control arm is a digitally constructed comparator group, generated using historical imaging and clinical data to simulate how a patient population would behave under standard of care, or if required, treatment-naïve conditions. At Qureight, we apply a machine learning-based approach to generate these synthetic arms, combining imaging and clinical biomarkers to create high-fidelity patient matches and model projected disease trajectories. In rare and progressive diseases like IPF, this approach offers both scientific rigor and ethical sensitivity. Traditional placebo-controlled trials can be difficult to justify when promising therapies are in play, and patient recruitment is often hindered by the risk of receiving no active treatment. Our synthetic arms provide an alternative that reduces reliance on placebo groups while preserving statistical integrity and regulatory viability. This enables faster trial execution, improved enrollment rates, and supports the design of adaptive trials. It also allows for smaller, more focused studies without sacrificing robustness—an especially valuable feature in rare disease settings. We presented data at the 2025 American Thoracic Society Congress from two independent Phase II IPF studies, in which Qureight’s synthetic control arms closely mirrored the behavior of real-world control groups across multiple endpoints. These results validate the methodology and mark a significant step toward regulatory-grade synthetic arms in pulmonary drug development. Ultimately, synthetic arms are not just a novel statistical strategy—they represent a step-change in trial design, accelerating the development of new therapies while upholding both ethical and scientific standards.
Q: Looking at the company’s trajectory since its formation in 2018, what were the foundational challenges you and the co-founders faced in building a company at the intersection of medicine and deep learning? What key milestones or moments have been most critical to the company’s progress and growth to date?
Walsh: From the outset, one of the core challenges in building Qureight was forging a meaningful bridge between two fundamentally different disciplines: clinical medicine and machine learning. These worlds operate on different timescales, with different standards of validation, language, and culture. Building a shared framework required us to earn trust across multiple fronts—pharma, academia, regulators—while simultaneously delivering tools that were both scientifically credible and scalable. It was not enough to build interesting models; we had to embed them in a framework that respected the regulatory and clinical expectations of modern trials. Key milestones that stand out include the first time our imaging biomarkers were implemented in an interventional trial, and the publication of our early validation studies in high-impact journals. These gave our platform credibility in the eyes of sponsors and investigators. Raising our Series A funding round also marked a critical inflection point and proof that the commercial world saw what we saw: a future where imaging biomarkers were central to trial strategy, not peripheral. The rapid expansion of our partnerships and the emergence of our LungAI suite signaled that we had moved beyond startup experimentation into real-world deployment.
Q: Qureight is currently focused on lung and heart diseases. What new projects or disease areas are in the pipeline for expansion, and what criteria do you use to select these new therapeutic areas? How do you foresee your technology being applied to other conditions where complex data analysis is a barrier to precision medicine?
Walsh: Our current focus is on fibrotic lung disease and pulmonary vascular disorders. This reflects areas where imaging plays a central role in disease tracking but has historically lacked quantification. From here, we’re expanding into heart failure—both preserved and reduced ejection fraction phenotypes—because these areas suffer from the same challenges: subjective assessment, complex multimodal data, and a need for clearer endpoints in trials. We prioritize new disease areas where there is a strong biological rationale, unmet clinical need, and fragmented data environments that machine learning can bring into alignment. We’re also actively exploring rare diseases and autoimmune conditions, especially where imaging intersects with other high-dimensional data streams. While our focus remains on supporting clinical trial design and regulatory strategy, the long-term vision is broader. There are specialist clinical centers that are increasingly data-rich but insight-poor. Our technology can close that gap. Over time, we believe structured imaging and AI-derived metrics will become core to precision medicine, not just in trials but in daily decision-making.
Q: In the broader context of precision medicine, how important do you believe Qureight’s technology will be in enabling a more personalized approach to patient care and treatment? Are there plans to move beyond supporting clinical trials to also provide tools for clinicians in their day-to-day practice?
Walsh: In the broader context of precision medicine, Qureight’s platform plays a critical role in precisely quantifying prognostic signals from HRCT—not in aggregate, but from each anatomical and functional compartment of the lung. We separate and quantify vascular, airway, and parenchymal signals, allowing us to track how different drugs act on specific biological targets. This level of precision isn’t just desirable, it’s essential for matching mechanism of action to measurable treatment effect. Think of it like getting a tailored suit: you wouldn’t settle for a single chest measurement and hope for the best. You’d want the tailor to take multiple detailed measurements to ensure a perfect fit. Precision medicine is no different. At Qureight, we take multiple, compartment-specific measurements of the lung to build a complete and personalized understanding of disease progression and drug effect. We believe this mechanistic dissection of imaging data will be central to the future of trial design, especially in diseases like IPF and pulmonary hypertension, where traditional endpoints like FVC are crude, liable to significant measurement variability, and often uninformative in early-phase studies. Right now, our focus is on maximizing the impact of this technology in drug development, where the unmet need is acute and where precision imaging biomarkers can accelerate timelines, de-risk programs, and sharpen regulatory discussions. Clinical trials are the pressure point—and it’s where we can have the biggest effect today.
Q: Given the rapid pace of innovation in AI, what is Qureight’s strategy for staying at the forefront of deep learning and data analytics? How do you ensure your technology remains proprietary and cutting-edge in a competitive landscape?
Walsh: Qureight’s strategy is built around a deep integration of clinical insight and technical excellence. Our core belief is that innovation in AI is only meaningful when grounded in biological relevance, medical credibility, and translational utility. To that end, we maintain an in-house team of clinical radiologists, data scientists, and machine learning engineers who collaborate closely on every model, from conception to deployment in trials. This enables us to develop task-specific, medically validated models rather than relying on generic AI tools. We invest continuously in internal R&D, with a strong publication track record that underpins our credibility and ensures that our models stand up to peer scrutiny. Our model suite is modular and scalable, allowing us to iterate quickly as new data types or endpoints become relevant. Importantly, we retain intellectual property over our analytics pipeline and design outputs with regulatory adoption in mind from the outset. Proprietary value also comes from our data partnerships. We focus on structured, high-quality, and longitudinal datasets—often drawn from real-world registries or carefully curated trial cohorts—which are essential for disease modelling. By aligning with leading academic centers and life science sponsors, we ensure our insights remain both cutting-edge and deeply embedded in the clinical research ecosystem. In a crowded space, our credibility, partnerships, and focus on trial-readiness distinguish us.
Damian Doherty has been in media and publishing for over 30 years, beginning at News Corporation. Damian has managed, edited, and launched life science titles in drug discovery and precision medicine. He was features editor of Drug Discovery World and founded the Precision Medicine Leaders Summit and the Journal of Precision Medicine. He edited AIMed magazine before launching Photo51Media, a platform for illuminating untold, compelling stories in precision healthcare. Damian joined Mary Ann Liebert in 2021 to help steer the new rebrand and relaunch of Clinical OMICS to Inside Precision Medicine.
