Bridging the Color Gap: AI Optical Scanner Tackles Hospital Bedsores

ROSH PINNA, Israel – November 04, 2026 — The intersection of healthcare equity and automation has rarely been as literal as it is at the patient bedside. For decades, the standard of care for detecting hospital-acquired pressure injuries—commonly known as bedsores—has relied on the human eye and a nurse's touch. But this analog approach carries a profound, systemic flaw: it routinely fails patients with darker skin tones.

Today, Israel-based medical device developer IR-MED Ltd. announced a significant clinical milestone that could shift the paradigm of wound care. The company revealed the publication of peer-reviewed interim Phase 2 results in the HCA Healthcare Journal of Medicine for its investigational Infra-Red Spectroscopy Scanner (IRSS), the technology underlying its PressureSafe platform. The machine-learning-enabled device demonstrated an 87.7% overall accuracy rate in distinguishing healthy intact tissue from early-stage pressure injuries and deep tissue injuries.

More importantly, the trial directly targeted the diagnostic color gap, with approximately 54% of participants representing darker skin phototypes (Fitzpatrick IV–VI). As healthcare systems increasingly look to artificial intelligence to solve operational inefficiencies and eliminate bias, IR-MED's optical scanner offers a compelling case study in how technology can augment human clinical judgment to deliver more equitable care.

Beyond the Eye: The Diagnostic Color Gap

To understand the business and patient-side impacts of this technology, one must first understand the clinical crisis it aims to solve. Pressure injuries often begin beneath visually intact skin. Traditional dermatological checks rely on identifying "non-blanchable erythema"—a localized redness that does not turn white when pressed.

However, in patients with high melanin levels, this crucial visual cue is frequently obscured. Pigmentation can mask the early inflammatory stages of tissue damage, leading to delayed or missed diagnoses. By the time a deep tissue injury (DTI) becomes visually apparent in a darker-skinned patient, it has often progressed to a severe stage, presenting as a purple or maroon discoloration or a blood-filled blister.

Undetected or late-stage pressure injuries result in catastrophic patient outcomes, including severe pain, systemic infections like sepsis, and significantly increased mortality rates. Financially, they represent a multibillion-dollar burden on the global healthcare system, with treatment costs escalating exponentially as the wound severity increases. Organizations such as the National Pressure Injury Advisory Panel (NPIAP) have long emphasized the urgent need for objective assessment tools that go beyond visual inspection, particularly to address racial disparities in wound care outcomes.

The Technology: Infrared Spectroscopy Meets Machine Learning

The investigational IRSS device underlying the PressureSafe platform bypasses the epidermal pigmentation barrier entirely. The handheld, battery-operated tool utilizes diffuse-reflectance spectroscopy across visible and near-infrared wavelengths. In approximately 400 milliseconds, it gathers bio-information from beneath the skin, assessing optical signals associated with underlying structural characteristics and tissue biomarkers indicative of pressure injuries.

This data is instantly processed by IR-MED's proprietary machine-learning classifier to provide a real-time, objective assessment. The technology is explicitly designed as a point-of-care decision-support tool—meaning it is intended to augment, not replace, the visual and tactile assessments performed by qualified healthcare professionals.

This approach distinguishes IR-MED from other players in the growing objective wound assessment market. While competitors have made substantial headway with FDA-cleared sub-epidermal moisture (SEM) scanners that detect localized edema, IR-MED's technology focuses on a broader range of physiological biomarkers via infrared spectroscopy. If successfully commercialized, this could offer clinicians a highly nuanced, rapid-fire tool that integrates seamlessly into routine acute-care workflows.

Promising Interim Data with Modest Cohorts

The data published in the HCA Healthcare Journal of Medicine (Volume 7, Issue 5) provides a promising, albeit early, look at the device's efficacy. The prospective, two-phase pilot study was conducted at two acute-care facilities in San Antonio, Texas: Methodist Hospital Metropolitan and Methodist Hospital Stone Oak.

Across both phases, the study included 26 participants contributing 219 anatomical-site observations and 1,314 individual device scans. The newly published interim Phase 2 analysis, however, focused on a smaller subset: nine participants, 65 anatomical sites, and 390 individual scans, which included ten sites clinically classified as pressure injuries by Certified Wound Ostomy Continence Nurses.

Using a post hoc balanced classification threshold, the interim Phase 2 analysis reported 87.7% overall accuracy, 90.0% sensitivity, and 87.3% specificity.

“These peer-reviewed results represent an important clinical milestone for IR-MED and provide independent validation of the potential of the infrared spectroscopy and machine-learning technology underlying our PressureSafe platform,” said Dr. Yaniv Cohen, Co-Founder and Chief Scientific Officer of IR-MED.

Cohen noted that the Phase 2 results represent a 6.5 percentage-point improvement in accuracy and a 7.6 percentage-point improvement in specificity compared to Phase 1, highlighting the algorithm's ability to learn and refine its diagnostic performance from high-quality clinical data. The Area Under the Receiver Operating Characteristic (AUC-ROC) curve—a robust measure of a diagnostic test's overall accuracy—increased to a highly impressive 0.959.

Dr. Mary Lee Potter, the study’s principal investigator and lead author, echoed this optimism. “Early-stage pressure injuries and deep tissue injuries can be difficult to identify, particularly when tissue damage develops beneath intact skin or when visual signs are less apparent," Potter stated. "The interim findings demonstrate promising agreement between the investigational infrared spectroscopy scanner underlying PressureSafe and assessments performed by experienced wound-care clinicians.”

Notably, no device-related safety events, adverse tissue reactions, or patient discomfort were identified during either study phase.

Navigating Limitations and the Road to Commercialization

While the interim results are undeniably strong, clinical researchers and biotech analysts must view them through the lens of early-stage trial limitations. The Phase 2 interim cohort of nine patients is a highly modest sample size. Furthermore, the reported metrics were achieved using a "post hoc balanced classification threshold." While an AUC-ROC of 0.959 indicates excellent overall discrimination independent of a threshold, sensitivity and specificity are highly dependent on where that threshold is set. Adjusting parameters after data collection is a standard practice in the early training phases of machine learning models, but it carries an inherent risk of overfitting the algorithm to a specific dataset. The true test of the PressureSafe platform will be its ability to maintain these high performance metrics on generalized, unseen data in much larger patient populations.

Additionally, industry observers will likely note the institutional synergy at play. The trial was conducted at Methodist Hospitals, which are part of the HCA Healthcare network, and the results were published in the HCA Healthcare Journal of Medicine. While the journal is peer-reviewed and the findings are transparent, this close ecosystem highlights IR-MED's strong institutional relationships—a factor that could prove highly advantageous for future clinical adoption, provided the technology clears regulatory hurdles.

For IR-MED, an OTC-listed company (IRME), the path forward requires substantial capital and rigorous clinical execution. Phase 2 enrollment and data collection remain ongoing under protocol IRM-02-1123, with completion currently expected by March 2027. This continued enrollment is critical for expanding the clinical dataset, improving the precision of performance estimates, and validating the scanner across a wider array of anatomical sites and skin-tone groups.

As healthcare systems face mounting pressure to deliver equitable, cost-effective care, the demand for objective, AI-driven diagnostic tools will only accelerate. If IR-MED can replicate its impressive interim results at scale, its 400-millisecond scan could soon become an essential weapon in the fight against a hidden, systemic healthcare crisis, proving that sometimes, the most profound innovations are the ones that help us see what the human eye cannot.