The scale of the data infrastructure behind the study is itself remarkable. Rather than relying on surveys or small clinical cohorts, the researchers accessed anonymised electronic health records spanning general practice, hospital episode statistics for admitted patient care and outpatient care, and mortality records, all linked at the individual level within a secure research environment. This allowed the team to calculate age- and sex-standardised prevalence for each of the 406 rare diseases and to compute recorded COVID-19 infection rates for the overall rare disease group, for each individual condition, and across both the six-group Office for National Statistics ethnicity classification and the more granular 19-group NHS classification.

The headline finding is that 1.05 million people in England, or 1.7 percent of the population studied, had at least one of the 406 rare diseases captured in the analysis. While that proportion may appear modest, it represents a substantial population whose healthcare needs are often fragmented across specialties and whose conditions may go unrecognised for years. By standardising prevalence estimates for age and sex, the researchers accounted for the demographic differences between ethnic groups, ensuring that observed variations reflect genuine differences in disease occurrence rather than differences in population structure.

Perhaps the most striking technical insight of the study concerns the way ethnic categories are constructed. When the researchers used broad aggregate categories, patterns of rare disease prevalence appeared relatively uniform. But when these categories were disaggregated into the finer-grained NHS classifications, substantial heterogeneity emerged that had been completely hidden. A vivid example is discoid lupus erythematosus, an autoimmune condition affecting the skin: its prevalence was markedly elevated among people recorded as Black Caribbean but not among those recorded as Black African, a distinction that vanishes entirely when both groups are folded into a single aggregate Black category. This finding has direct implications for how health systems collect and report ethnicity data, suggesting that coarse classifications can mask clinically meaningful differences in disease distribution.

The COVID-19 component of the analysis adds a pandemic-era dimension to the picture. Recorded COVID-19 infection was higher among people with rare diseases than in the rest of the population, at 32.8 percent compared with 29.0 percent. This elevation could reflect a combination of factors, including greater contact with healthcare settings where infections were detected, potentially increased susceptibility related to underlying conditions, and differences in testing behaviour. Because the study measured recorded infections rather than confirmed biological incidence, some caution is warranted in interpreting the absolute rates, but the relative patterns remain informative.

Within the rare disease group, infection rates varied considerably by ethnicity. People in the Pakistani group had a recorded infection rate of 27.7 percent, and those in the Black African group 23.8 percent, compared with 33.7 percent in the White British group. Infection rates also varied by specific rare disease, ranging from 24.6 to 41.2 percent among the 25 most common conditions in the dataset. The observation that the highest recorded rates appeared in the ONS White and NHS White British groups is noteworthy, as it runs counter to some earlier pandemic-era findings that reported higher recorded infection rates in several minority ethnic groups, and it underscores how the intersection of rare disease status and ethnicity produces patterns that neither factor alone would predict.

The study’s methodology deserves attention for how it handled the complexity of rare disease identification. Rare diseases are defined individually by low prevalence, often affecting fewer than one in 2,000 people, which means that even a dataset of 62.5 million people yields small counts for many individual conditions. The researchers used coded diagnostic criteria drawn from ICD-10 and SNOMED CT terminologies to identify cases across primary care, hospital, and mortality records, and they reported confidence intervals alongside their prevalence estimates to convey the precision of each figure. The breadth of the approach, covering 406 distinct conditions, means the resulting dataset functions as a baseline reference that clinicians, commissioners, and researchers can consult when assessing whether specific communities are being diagnosed fairly.

The implications for health equity are significant. If the prevalence of certain rare diseases is genuinely higher in particular ethnic groups, whether because of genetic ancestry, environmental exposures, or a combination of factors, then diagnostic pathways, screening programmes, and specialist services need to be designed with that distribution in mind. Conversely, if prevalence appears lower in some groups, this may signal underdiagnosis rather than true protection, a possibility that the authors suggest their findings can help interrogate. The study’s authors frame their work as providing a baseline to support earlier, fairer diagnosis and more targeted care, positioning the national map as a tool for identifying where diagnostic effort should be concentrated.

There are also important caveats that shape how the findings should be read. The cross-sectional design captures a snapshot rather than tracking individuals over time, so it cannot establish when diagnoses were made or how incidence has changed. Recorded infection status depends on testing uptake and recording practices, which themselves varied across the pandemic and across communities. Ethnicity recording in health records is imperfect, and some individuals will have missing or inconsistently coded ethnicity. The study also covers England specifically, and healthcare system differences mean the findings may not generalise directly to other countries, though the methodological framework is broadly transferable.

Nevertheless, the study stands as a landmark in the use of national-scale linked health records to address questions that smaller datasets simply cannot answer. By demonstrating that rare disease prevalence varies substantially across ethnic groups and that much of this variation is invisible within broad categories, the research makes a compelling case for finer-grained ethnicity data collection throughout health systems. As genomic medicine, newborn screening, and personalised therapies expand the possibilities for rare disease care, ensuring that these advances reach all communities equitably will depend on exactly the kind of evidence this study provides: a detailed, ethnicity-resolved accounting of who has which conditions, and how they fared during one of the most disruptive public health emergencies in recent history.

Subject of Research: Prevalence of 406 rare diseases by ethnicity and recorded COVID-19 infection rates in England

Article Title: Prevalence of 406 rare diseases by ethnicity and recorded COVID-19 infection rates among 62.5 million people in England

Article References: Gu, Q., Hasheminasab, S. A., Pineda-Moncusí, M., Chilala, C., Thygesen, J. H., Wu, H., Khalid, S., Coombe, A., & CVD-COVID-UK/COVID-IMPACT Consortium (2026). Prevalence of 406 rare diseases by ethnicity and recorded COVID-19 infection rates among 62.5 million people in England. Scientific Reports. https://doi.org/10.1038/s41598-026-70531-4

Image Credits: AI Generated

DOI: 10.1038/s41598-026-70531-4

Keywords: rare diseases, ethnicity, COVID-19, electronic health records, prevalence, health inequalities, NHS England, epidemiology, population health, health equity, discoid lupus erythematosus, data linkage

The research, led by Pordil Khan and Abdullah Abdullah of Khyber Medical College in Peshawar, Pakistan, together with collaborators at institutions in Pakistan and Afghanistan, set out to address a gap that has long frustrated dermatopathology, particularly in resource-constrained settings. Machine learning studies on skin lesions have tended to focus on dermatoscopic images of pigmented lesions and skin cancer, where large public datasets exist. Multi-class classification of inflammatory dermatoses from histopathological images, by contrast, has remained comparatively unexplored, even though these conditions are common, diagnostically slippery, and heavily dependent on scarce expert interpretation. The team assembled a balanced dataset of 96 original histopathological images per class, drawn from biopsy specimens collected as part of routine clinical care at Khyber Medical College, with ethics approval from the institution’s review board and written informed consent from patients at the time of biopsy.

Because deep learning models are notoriously data-hungry and 96 images per category is a small foundation, the researchers turned to augmentation, a technique that programmatically expands a training set by applying transformations such as rotations, flips, and other pixel-level adjustments that preserve the diagnostic content of the image. Using the Albumentations library, they applied this process offline and exclusively to the training data, never to the validation images, expanding each class to 486 images. This distinction matters: augmenting only the training set ensures that the models are evaluated on genuine, untouched biopsy images rather than on variations of pictures they have already seen, which would inflate performance estimates and tell clinicians little about real-world behavior.

At the heart of the study was a comparison of four pre-trained convolutional neural networks, each fine-tuned for the four-class task using transfer learning, the strategy of starting from a network already trained on millions of general images and adapting it to a specialized medical problem. ResNet50, a 50-layer architecture whose residual connections allow very deep networks to train stably, emerged as the clear leader, reaching 96.67 percent accuracy on the validation cohort with a micro-average area under the receiver operating characteristic curve of 0.996, a figure indicating near-perfect separation between the disease classes across the full range of decision thresholds. EfficientNet-B0, a compact architecture that scales network depth, width, and image resolution in a balanced way, followed at 95.00 percent accuracy with a micro-average AUROC of 0.990. MobileNetV2, designed for lightweight mobile deployment, managed 85.00 percent, while the older VGG19 architecture trailed at 78.33 percent, a result that reflects how quickly convolutional network design has advanced since VGG’s introduction.

The evaluation protocol deserves attention because it was designed to avoid one of the most common pitfalls in medical image analysis. The models were assessed using an approximately 85:15 patient-level training-validation split, meaning that all images from a single patient were kept on one side of the divide. The validation cohort consisted of 60 original images from 20 patients. Patient-level splitting prevents a model from effectively memorizing the idiosyncrasies of one person’s tissue and then being tested on other slides from the same person, a leakage problem that has produced deceptively high accuracy figures in many published AI studies. Performance was measured with a battery of standard metrics, including accuracy, precision, recall, F1-score, AUROC, the area under the precision-recall curve, and confusion matrices that lay out exactly where the models stumbled.

Those confusion matrices revealed a telling pattern. For ResNet50, every one of the 15 validation images of eczema and every one of the 15 images of lupus vulgaris was classified correctly, while discoid lupus erythematosus and dermatitis herpetiformis each accounted for a single misclassified image among their 15 validation slides. In other words, only two of the 60 validation images were assigned to the wrong disease category, and the errors occurred between conditions that pathologists themselves find difficult to separate. This kind of granular error analysis is far more informative for clinicians than a single headline accuracy number, because it identifies which diagnostic boundaries the algorithm has genuinely mastered and which remain contested territory.

Perhaps the most consequential element of the work, at least for clinical acceptance, is its use of explainability techniques. Deep networks are often criticized as black boxes: they produce a label but no rationale, and a pathologist asked to trust an opaque verdict on a biopsy is likely to decline. The researchers applied Grad-CAM++, a gradient-based visualization method that generates heatmaps highlighting the image regions most responsible for a model’s prediction. When these heatmaps are overlaid on the original histopathological image, they show whether the network is attending to diagnostically meaningful structures, such as the characteristic neutrophilic deposits at the dermal papillae seen in dermatitis herpetiformis or the granulomatous inflammation of lupus vulgaris, or whether it is latching onto irrelevant artifacts like staining variation or cutting marks. Qualitative visualization of this kind turns the model from an oracle into a colleague whose reasoning can be inspected, checked, and, when necessary, overruled.

To move the work beyond a laboratory exercise, the team also built a web-based prototype as an AI-assisted research tool for supportive image classification. Such prototypes are a familiar bridge in the medical AI literature: they demonstrate that a trained model can be packaged into an interface that a clinician or researcher could actually use, uploading a histopathological image and receiving a classification alongside its explanation. The authors are careful, appropriately, to frame this as a research prototype rather than a diagnostic device, and the study’s own limitations section is candid about why. The dataset was small and came from a single center, and there was no external validation on images from other hospitals, scanners, or staining protocols, all of which are known to degrade model performance when conditions differ from the training environment.

Those caveats do not diminish the significance of the demonstration; they define the roadmap. The authors explicitly call for future studies to evaluate larger, multi-center datasets and to incorporate additional clinical information, such as patient demographics and laboratory findings, which could be fused with image features to improve robustness. The distinction between feasibility and clinical applicability is one that the field of medical imaging AI has had to learn repeatedly, and this study draws it honestly. What the results establish is that the diagnostic signal separating these four inflammatory dermatoses is present in histopathological images at a strength that modern convolutional networks can extract, even from a modestly sized, single-institution dataset, provided that training is handled carefully with augmentation, transfer learning, and patient-level validation.

The broader implications reach well beyond dermatopathology. In many parts of the world, expert histopathologists are scarce, and diagnostic delays for conditions like cutaneous tuberculosis or lupus can be measured in months or years. A tool that offers a rapid, explainable second opinion on a biopsy image, flagging the most likely diagnosis and pointing to the tissue regions that support it, could help prioritize cases, reduce errors, and support training of junior pathologists. The finding that ResNet50 and EfficientNet-B0, both freely available architectures, achieved AUROC values above 0.99 on this task suggests that the barrier is not the sophistication of the model but the availability of well-curated, diverse data. If subsequent multi-center studies confirm these results, explainable deep learning could become a routine assistant at the microscope, transforming one of pathology’s most subjective judgment calls into a transparent, quantifiable, and auditable process, and bringing that capability first to the settings where expert eyes are hardest to find.

Subject of Research: Explainable deep learning classification of inflammatory skin diseases from histopathological images

Article Title: Explainable deep learning-based multi-class classification of inflammatory dermatoses from histopathological images

Article References: Khan, P., Abdullah, A., Ahmed, M., Ullah, Q. M. F., Umer, H., Qasim, M., Aizad, G., Gandapur, T. K., & Talha, M. (2026). Explainable deep learning-based multi-class classification of inflammatory dermatoses from histopathological images. BMC Medical Imaging. https://doi.org/10.1186/s12880-026-02897-w

Image Credits: AI Generated

DOI: 10.1186/s12880-026-02897-w

Keywords: deep learning, histopathology, dermatitis herpetiformis, discoid lupus erythematosus, eczema, lupus vulgaris, ResNet50, Grad-CAM++, transfer learning, dermatopathology, medical imaging AI, Pakistan