Parminder Singh Reel

Background: Early identification of individuals at risk of dementia is essential for preventive care and timely enrolment into disease-modifying interventions. However, most existing prediction approaches rely on invasive, costly, or research-only biomarkers that are not scalable within public healthcare systems. Routinely acquired National Health Service (NHS) brain magnetic resonance imaging (MRI) scans, when linked with electronic health records, represent a widely available and privacy-preserving resource for population-level dementia risk stratification. A key challenge for clinical translation is ensuring that machine-learning predictions are reliable, interpretable, and safe to apply, particularly when models are used years before clinical diagnosis. Methods: We conducted a retrospective case–control study entirely within a secure NHS Trusted Research Environment using routine T1-weighted brain MRI scans linked to electronic health records from Tayside and Fife, Scotland. The study included 518 participants: 259 individuals who subsequently developed dementia and 259 age- and sex-matched controls. Structural brain features were derived from MRI data and analysed using a support-vector-machine classifier with nested cross-validation to minimise overfitting. Prediction confidence was quantified using distance-from-hyperplane (DFH) calibration, enabling stratification of model outputs by certainty. Primary outcomes were classification accuracy and area under the receiver-operating-characteristic curve (AUC). Secondary analyses examined DFH-stratified performance and the relationship between prediction accuracy and time from scan to first recorded dementia diagnosis. Results: The model predicted future dementia up to five years before first recorded NHS diagnosis with an AUC of 0.71, a performance consistent with real-world clinical imaging rather than research-optimised datasets. Model sensitivity increased for scans acquired closer to diagnosis, indicating stronger predictive signal as disease onset approached. Confidence-based stratification identified a high-confidence subgroup comprising approximately 35% of scans, within which prediction accuracy increased to around 80%. Performance was consistent across heterogeneous routine NHS scanners and imaging protocols, demonstrating robustness and generalisability to real-world clinical data rather than research-optimised acquisitions. Conclusion: Routinely collected NHS brain MRI data can be used to predict future dementia several years before clinical diagnosis. Incorporating confidence calibration transforms a conventional machine-learning classifier into a safety-aware and clinically interpretable framework by enabling selective use of high-certainty predictions. This approach supports scalable early detection, population-level risk stratification, and targeted recruitment into preventive or disease-modifying clinical trials, with clear potential for integration into public health systems.

Importance: Early identification of dementia risk is essential for preventive care and timely enrolment into disease-modifying interventions. Current approaches rely on costly, invasive, or research-only methods not feasible at scale within public health systems. Objective: To test whether routinely acquired NHS brain MRI scans can be used to predict future dementia diagnosis and whether confidence-based stratification improves prediction reliability and clinical interpretability. Design, Setting, and Participants: Retrospective case-control study conducted entirely within a secure NHS Trusted Research Environment. Routine T1-weighted MRI brain scans were linked with electronic health records for participants from Tayside and Fife, Scotland. The study included 259 individuals who subsequently developed dementia and 259 age- and sex-matched controls. Data were processed and modelled between January and June 2025. Exposure: MRI-derived structural brain features analysed using a support-vector-machine model with nested cross-validation and distance-from-hyperplane (DFH) confidence calibration. Main Outcomes and Measures: Primary outcomes were prediction accuracy and area under the receiver-operating-characteristic curve (AUC). Secondary analyses assessed DFH-stratified performance and time from scan to clinical diagnosis. Results: Confidence-based filtering identified a high-confidence subgroup (≈35% of scans) with ≈80% accuracy. Overall, the model predicted future dementia up to five years before first recorded NHS diagnosis, achieving 66.8% accuracy (AUC = 0.71). Model sensitivity increased for shorter time-to-diagnosis intervals. Analyses were generalisable across heterogeneous routine NHS scanners and datasets. Conclusions and Relevance: This study provides, to our knowledge, the first demonstration that routinely collected NHS MRI data can predict future dementia years before clinical diagnosis. Incorporating confidence calibration transforms a standard classifier into a safety-aware, clinically interpretable framework, supporting scalable early detection, risk stratification, and recruitment to preventive or disease-modifying trials across population health systems. Key Points: Question: Can routinely acquired NHS magnetic resonance imaging (MRI) brain scans and linked health records be used to predict future dementia, with confidence calibration suitable for clinical and population-level use? Findings: In this retrospective case-control study of 518 individuals, a support-vector-machine model trained on routine NHS MRI data predicted future dementia up to five years before diagnosis. Confidence-based stratification identified a high-confidence subgroup (∼35% of scans) with ∼80% accuracy. Meaning: Routine NHS imaging data can be transformed into a population-scale, privacy-preserving early detection framework, where confidence calibration allows safe, clinically interpretable application for preventive care and research recruitment.

We present the design and implementation of a data curation framework to generate a large-scale clinical brain imaging dataset suitable for artificial intelligence (AI) enabled image analysis, accessible through the Brain Health Data (BHD) initiative. The raw data accessible through the BHD includes approximately 417K magnetic resonance imaging (MRI) and 846K computerized tomography (CT) head scans, linked electronic health records (EHRs), and associated free-text imaging reports from clinical practice between 2010 and 2018 in Scotland, totally exceeding 185 TB storage of brain imaging and associated data. We present the work curating the dataset and the strengths of the BHD, including clinical relevance thanks to its unprecedented scale, population-wide representativeness of a national free-at-the-point-of-service healthcare, long-term follow-up to neurodegenerative disease, and real-world variability. We discuss challenges and lessons learnt in developing the framework to curate the data initially available, including the time needed to obtain relevant permissions, the need for easily accessible, secure, responsive and affordable computational environments, the variability and inconsistencies of clinical data and records, and the challenge of extracting linked clinical data and images at scale, among others. This resource will be crucial for clinical research, fostering the development of personalized medicine approaches, and fast-tracking the implementation of AI models in clinical workflows. We encourage the use of the BHD data through a streamlined application to the Data Research and Innovation Service (eDRIS) of Public Health Scotland (PHS).

Keywords: MRI, Imaging Sequences, Ultrasound, Mammography, CT, Angiography, Conventional Radiography Published under a CC BY 4.0 license. See also the commentary by Whitman and Vining in this issue.

Headache is one of the commonest complaints that doctors need to address in clinical settings. The genetic mechanisms of different types of headache are not well understood while it has been suggested that self-reported headache and self-reported migraine were genetically correlated. In this study, we performed a meta-analysis of genome-wide association studies (GWAS) on the self-reported headache phenotype from the UK Biobank and the self-reported migraine phenotype from the 23andMe using the Unified Score-based Association Test (metaUSAT) software for genetically correlated phenotypes (N = 397,385). We identified 38 loci for headaches, of which 34 loci have been reported before and four loci were newly suggested. The LDL receptor related protein 1 (LRP1)—Signal Transducer and Activator of Transcription 6 (STAT6)—Short chainDehydrogenase/Reductase family 9C member 7 (SDR9C7) region in chromosome 12 was the most significantly associated locus with a leading p value of 1.24 × 10^–62 of rs11172113. The One Cut homeobox 2 (ONECUT2) gene locus in chromosome 18 was the strongest signal among the four new loci with a p value of 1.29 × 10^–9 of rs673939. Our study demonstrated that the genetically correlated phenotypes of self-reported headache and self-reported migraine can be meta-analysed together in theory and in practice to boost study power to identify more variants for headaches. This study has paved way for a large GWAS meta-analysis involving cohorts of different while genetically correlated headache phenotypes.

Hypertension is a major global health problem with high prevalence and complex associated health risks. Primary hypertension (PHT) is most common and the reasons behind primary hypertension are largely unknown. Endocrine hypertension (EHT) is another complex form of hypertension with an estimated prevalence varying from 3 to 20% depending on the population studied. It occurs due to underlying conditions associated with hormonal excess mainly related to adrenal tumours and sub-categorised: primary aldosteronism (PA), Cushing’s syndrome (CS), pheochromocytoma or functional paraganglioma (PPGL). Endocrine hypertension is often misdiagnosed as primary hypertension, causing delays in treatment for the underlying condition, reduced quality of life, and costly antihypertensive treatment that is often ineffective. This study systematically used targeted metabolomics and high-throughput machine learning methods to predict the key biomarkers in classifying and distinguishing the various subtypes of endocrine and primary hypertension. The trained models successfully classified CS from PHT and EHT from PHT with 92% specificity on the test set. The most prominent targeted metabolites and metabolite ratios for hypertension identification for different disease comparisons were C18:1, C18:2, and Orn/Arg. Sex was identified as an important feature in CS vs. PHT classification.

Despite considerable morbidity and mortality, numerous cases of endocrine hypertension (EHT) forms, including primary aldosteronism (PA), pheochromocytoma and functional paraganglioma (PPGL), and Cushing’s syndrome (CS), remain undetected. We aimed to establish signatures for the different forms of EHT, investigate potentially confounding effects and establish unbiased disease biomarkers. Plasma samples were obtained from 13 biobanks across seven countries and analyzed using untargeted NMR metabolomics. We compared unstratified samples of 106 PHT patients to 231 EHT patients, including 104 PA, 94 PPGL and 33 CS patients. Spectra were subjected to a multivariate statistical comparison of PHT to EHT forms and the associated signatures were obtained. Three approaches were applied to investigate and correct confounding effects. Though we found signatures that could separate PHT from EHT forms, there were also key similarities with the signatures of sample center of origin and sample age. The study design restricted the applicability of the corrections employed. With the samples that were available, no biomarkers for PHT vs. EHT could be identified. The complexity of the confounding effects, evidenced by their robustness to correction approaches, highlighted the need for a consensus on how to deal with variabilities probably attributed to preanalytical factors in retrospective, multicenter metabolomics studies.

Correlations between pain phenotypes and psychiatric traits such as depression and the personality trait of neuroticism are not fully understood. In this study, we estimated the genetic correlations of eight pain phenotypes (defined by the UK Biobank, n = 151,922–226,683) with depressive symptoms, major depressive disorders and neuroticism using the the cross-trait linkage disequilibrium score regression (LDSC) method integrated in the LD Hub. We also used the LDSC software to calculate the genetic correlations among pain phenotypes. All pain phenotypes, except hip pain and knee pain, had significant and positive genetic correlations with depressive symptoms, major depressive disorders and neuroticism. All pain phenotypes were heritable, with pain all over the body showing the highest heritability (h² = 0.31, standard error = 0.072). Many pain phenotypes had positive and significant genetic correlations with each other indicating shared genetic mechanisms. Our results suggest that pain, neuroticism and depression share partially overlapping genetic risk factors.

Background: The Health Informatics Centre at the University of Dundee provides a service to securely host clinical datasets and extract relevant data for anonymized cohorts to researchers to enable them to answer key research questions. As is common in research using routine healthcare data, the service was historically delivered using ad-hoc processes resulting in the slow provision of data whose provenance was often hidden to the researchers using it. This paper describes the development and evaluation of the Research Data Management Platform (RDMP): an open source tool to load, manage, clean, and curate longitudinal healthcare data for research and provide reproducible and updateable datasets for defined cohorts to researchers. Results: Between 2013 and 2017, RDMP tool implementation tripled the productivity of data analysts producing data releases for researchers from 7.1 to 25.3 per month and reduced the error rate from 12.7% to 3.1%. The effort on data management reduced from a mean of 24.6 to 3.0 hours per data release. The waiting time for researchers to receive data after agreeing a specification reduced from approximately 6 months to less than 1 week. The software is scalable and currently manages 163 datasets. A total 1,321 data extracts for research have been produced, with the largest extract linking data from 70 different datasets. Conclusions: The tools and processes that encompass the RDMP not only fulfil the research data management requirements of researchers but also support the seamless collaboration of data cleaning, data transformation, data summarization and data quality assessment activities by different research groups.