Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Ahmed M. Salih has a total of 104 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Advancing Heart Ageing in Ischemic Heart Disease: Insights from CMR and ECG Analysis

    by Ahmed M Salih, Elisa Rauseo, Ilaria Boscolo Galazzo, Esmeralda Ruiz Pujadas, Víctor M. Campello, Karim Lekadir, Nay Aung, Gregory G. Slabaugh, Ghaith Sharaf Dabbagh, Anwar A Chahal, et al.

  • Open Access

    Explainable Artificial Intelligence for the Detection and Characterisation of Stage B Heart Failure

    by Ahmed M Salih, Emer Brady, Ranjit Arnold, Gaurav Gulsin, Huiyu Zhou, Anvesha Singh, Gerry McCann

    Background: Stage B heart failure is characterised by asymptomatic structural or functional cardiac abnormalities. Identifying individuals at this stage of diseases is clinically important, as early detection may enable the use of targeted interventions to prevent progression to symptomatic heart failure. Explainable artificial intelligence (XAI) may facilitate early detection, transparent risk stratification, and appropriate selection of clinically actionable intervention to prevent disease progression. Objective: To review the use of XAI methods in detecting and characterising stage B heart failure. Methods: A literature search of Web of Science, Scopus, and PubMed was conducted on 27 March 2026. Studies were included if they applied AI with XAI techniques to stage B heart failure. After screening, 20 studies were included. Data on modalities, outcomes, demographic reporting, and XAI methods were extracted and synthesised. Results: SHAP was the most commonly used explainable AI method, followed by LIME, saliency maps, and Grad-CAM; however, XAI adoption remained inconsistent, with several studies relying on limited or ad hoc interpretability approaches. Importantly, none of the studies generated or compared explanations across sex or ethnic subgroups, despite evidence of subgroup-specific differences in disease burden. Furthermore, evaluation of XAI outputs was often lacking: some studies did not assess explanations at all, while others relied solely on literature-based comparisons, which may introduce bias and subjectivity. Together, these limitations indicate that explainability was not systematically validated or leveraged to support fair and generalisable clinical inference. Conclusion: XAI shows promise for improving transparency in Stage B heart failure identification, but current implementations remain limited. Major gaps include insufficient consideration of sex and ethnicity, lack of subgroup-specific explainability analyses, inconsistent XAI evaluation, and limited external validation, all of which constrain generalisability and clinical adoption.

  • Top Cited Papers
    Open Access

    Clinical Phenotypes in Hypertension: A Data-Driven Approach to Risk Stratification

    by Elisa Rauseo, Ahmed M. Salih, Jackie Cooper, Musa Abdulkareem, Christopher R.S. Banerji, Sucharitha Chadalavada, Hafiz Naderi, Patricia B Munroe, Anthony Mathur, Nay Aung, et al.

    BACKGROUND: Hypertension is a major contributor to cardiovascular morbidity and mortality. Its heterogeneity complicates risk stratification. Unsupervised machine learning can uncover risk profiles and refine preventative strategies. This study applied a data-driven approach to identify clinical phenotypes of hypertension, examine their associations with cardiovascular imaging characteristics and adverse outcomes, and assess the mediating role of cardiac imaging features in these associations. METHODS: Fourteen thousand eight hundred forty UK Biobank participants with diagnosed hypertension and cardiovascular magnetic resonance imaging were analyzed. K-means clustering was applied to 77 clinical variables. Associations with incident heart failure, atrial fibrillation, atherosclerotic events, all-cause mortality, and major adverse cardiovascular events were examined and adjusted for cardiovascular risk factors. Mediation analyses assessed the role of cardiovascular imaging features in the association between clusters and outcomes. RESULTS: Three clusters emerged. Cluster 1, predominantly female with the most favorable metabolic profile, had the lowest risk. Cluster 2, predominantly male with the highest atherosclerosis burden, carried the greatest risk for all adverse events, independent of cardiovascular risk factors. They showed severe cardiac remodeling, impaired cardiac mechanisms, and global left atrial dysfunction. Cluster 3 had a profile resembling metabolic syndrome, with moderate risk for atrial fibrillation and all-cause death (hazard ratio, 1.65 and 1.58; P <0.05). Although in cluster 2 the risk was largely mediated by left ventricular hypertrophy, in cluster 3 its role was attenuated and more evenly balanced with left atrial dysfunction. CONCLUSIONS: Clustering analysis identified distinct hypertension phenotypes with specific risk profiles, suggesting potential for improved stratification and more tailored treatment approaches.


  • Open Access

    Explainability Challenges in Medical AI: The Conceptual Hidden Cost of Machine Learning Preprocessing

    by Ahmed M. Salih

    Background: Data preprocessing is a significant step in machine learning to improve the performance of the model and decreases the running time. This might include dealing with missing values, outliers' removal, data augmentation, dimensionality reduction and handling the confounding variables. Objective: This commentary explores the common preprocessing steps used in medical machine learning and highlights their conceptual hidden cost including reduced model explainability and clinical interpretability. This commentary focuses on conceptual rather than empirical implications of these preprocessing decisions. Methods: Literature review was undertaken to explore the data preprocessing steps in machine learning. Results: Although it is found the preprocessing steps improve the accuracy of the model, but they might block new findings and hinder the explainability of the model if they are not carefully considered especially in medicine. We identify key risks such as bias introduction and over simplification and outline mitigation strategies. Conclusion: The novelty of this work lies in systematically connecting preprocessing practices with explainability challenges in healthcare artificial intelligence and suggesting approaches to balance performance with explainability. This highlights the need for careful design of preprocessing pipelines in medical artificial intelligence systems to ensure both reliable predictions and clinical trust.

  • Open Access

    From GWAS to Causal Inference: A Beginner’s Guide to Mendelian Randomization with Code Examples

    by Ahmed M Salih, Roman Roy, Yuhe Wang, Irene Treccani, Andre Altmann, Zahra Raisi-Estabragh, Gloria Menegaz

    Background: Mendelian randomization (MR) is a powerful approach for assessing causal relationships between risk factors and health outcomes using genetic variants as instrumental variables (IVs). The increasing availability of large genome-wide association study (GWAS) summary statistics from resources such as UK Biobank, FinnGen, and other population-based cohorts has made MR analyses more accessible than ever. However, many available guidelines and tutorials remain highly technical, requiring advanced knowledge of statistical genetics and R programming. Objective: This paper aims to provide a clear, step-by-step guide for conducting MR analyses using GWAS summary statistics, designed specifically for non-technical researchers. Methods: We outline a structured workflow covering key stages of MR analysis, including dataset selection, quality control, IVs selection, harmonization, and causal estimation. The workflow integrates online tools for quality control and demonstrates the use of commonly applied R packages such as TwoSampleMR. Each step is illustrated with example code and practical guidance to promote reproducibility. Results and Conclusion: The proposed workflow supports the process of conducting MR analyses, bridging the gap between theoretical guidelines and hands-on implementation. By offering an accessible and reproducible framework, this tutorial aims to help applied researchers, clinicians, and early-career scientists confidently perform MR analyses and interpret causal findings using publicly available GWAS summary data.

  • The Marriage of Neurotechnologies and Artificial Intelligence: Ethical, regulatory, and technological aspects

    by Carlo Chiurco, Andrea Favaro, Silvia Francesca Storti, Lorenza Brusini, Ahmed M. Salih, Ilaria Boscolo Galazzo, Sergey Plis, Gloria Menegaz

    The dual concepts of neurotechnology and artificial intelligence (AI) form an intriguing but also potentially explosive mixture because of its many ethical and legal implications. The advent of AI and the progress in neurotechnologies are reshaping the landscape not only in all scientific fields but also in everyday life both individually and collectively, ushering in a new era where the centrality, integrity and identity of humans is no longer a fact. Such tumultuous progress has implications at all levels, individual, societal, economical and political. Without the pretension of exploring the whole set of relevant aspects, we aim at providing a multi-disciplinary view on the main ethical, legal and societal issues stemming from neurotechnology and AI, by assessing them using keywords like trustworthiness, fairness, awareness, security, and privacy. In this paper, we propose an overview on the current scenario, taking a philosophical perspective in the light of ethics, and boiling it down to aspects closely related to the technological developments and the regulatory measures that are currently in-place and called for.


  • Open Access

    Explainable AI: A Practical Guide for Non-Technical Users

    by Ahmed M Salih, Steffen E Petersen

    As Artificial Intelligence (AI) systems become more embedded in everyday decision-making, the need for transparency and interpretability is increasingly urgent especially for non-technical users such as clinicians, educators, managers, and policymakers. Explainable AI (XAI) refers to a set of tools and techniques designed to make AI models more understandable, enabling users to interpret, trust, and effectively act upon model predictions. However, existing XAI resources often assume significant technical knowledge and fail to address practical considerations relevant to real-world applications. This paper provides a non-technical guide to how effectively use XAI. It introduces the concept of XAI within the broader AI development pipeline and explores how decisions made throughout from data collection and pre-processing to model choice and validation can influence the quality of explanations. Key issues such as data quality, pre-processing, confounding variables, model performance, and sample size are discussed in the context of their impact on interpretability. It also presents an overview of explanation techniques (e.g., SHAP, LIME) and guidance on evaluating the usefulness and reliability of these explanations. Finally, it emphasizes the need for thoughtful evaluation and cautions against over-reliance on explanations from poorly performing models. By the end, readers will be equipped to select appropriate XAI methods, interpret explanations critically, and integrate explainability into their decision-making processes even without a technical background.

  • Artificial intelligence in cardiovascular imaging: risks, mitigations and the path to safe implementation

    by James P Howard, Qiang Zhang, Ahmed M Salih, Steffen E Petersen, Karim Lekadir, Zahra Raisi-Estabragh

    Artificial intelligence (AI) is rapidly transforming cardiovascular imaging by automating tasks such as image segmentation, feature extraction, and risk prediction - leading to significant improvements in diagnostic precision and efficiency. However, the integration of AI into clinical workflows comes with critical risks that must be addressed to ensure safe and reliable patient care. This review explores the technical, clinical, and ethical challenges of AI in cardiovascular imaging, particularly highlighting the risks of model errors, data drift and inappropriate usage. We also examine concerns about explainability, the potential for deskilling of healthcare professionals, generalisability across diverse populations, and accountability in AI implementation. We present real-world examples of where these risks have been realised, along with attempts at mitigations, including the adoption of explainable AI techniques, rigorous validation frameworks to ensure fairness and broad applicability, continuous performance monitoring, and transparency at every stage of model development and deployment. The successful adoption of AI in cardiovascular imaging relies on striking a balance between innovation and the need for ethical and legal safeguards. Achieving this requires collaborative efforts between clinicians, data scientists, patients and regulators. Evaluating and addressing these challenges is essential for responsible AI implementation and advancing patient care while maintaining high safety standards.


  • Open Access

    Clinical phenotypes in hypertension: a data-driven approach to risk stratification and outcome prediction

    by Elisa Rauseo, Ahmed M. Salih, Jackie Cooper, Musa Abdulkareem, Christopher R.S. Banerji, Sucharitha Chadalavada, Hafiz Naderi, Patricia B. Munroe, Anthony Mathur, Nay Aung, et al.

    Background: Hypertension (HTN) is a major contributor to cardiovascular (CV) morbidity and mortality. Its heterogeneity complicates risk stratification. Unsupervised machine learning can identify risk profiles and refine preventative strategies. Objectives: This study applies clustering analysis to clinical data to identify HTN phenogroups and their link with CV abnormalities and outcomes. Methods: 14,840 UK Biobank participants with a diagnosis of HTN who underwent cardiovascular magnetic resonance (CMR) were analysed. K-means clustering was applied to 77 clinical variables. Associations with incident HF, atrial fibrillation (AF), atherosclerotic events, all-cause mortality, and major adverse cardiovascular events (MACE) were examined. Mediation analysis assessed the role of CV imaging metrics in risk stratification. Results: Three clusters emerged. Cluster 1, predominantly female with the most favourable metabolic profile, had the lowest risk. Cluster 2, predominantly male with the highest atherosclerosis burden, carried the greatest risk, particularly for AF and HF (Hazard ratio [HR] 1.80 and 1.85; p <0.005). This group showed the most severe cardiac remodelling, impaired cardiac mechanisms, and global left atrial (LA) dysfunction, approaching the HF myocardial substrate. Cluster 3 had an intermediate risk profile resembling metabolic syndrome, with moderate MACE risk and a higher susceptibility to AF (HR 1.30 and 1.61; p <0.05). While in cluster 2 the risk was largely mediated by LV remodelling, in cluster 3, its role was attenuated and more evenly balanced with LA dysfunction. Conclusions: Clustering analysis revealed distinct HTN phenotypes with unique clinical and imaging profiles, enhancing risk stratification and supporting more individualised treatment strategies.


  • Open Access

    Cardiovascular magnetic resonance reference ranges for cardiac function and structure and recommendations for grading severity: the Healthy Hearts Consortium

    by Liliana E. Szabo, Celeste McCracken, Dorina Gabriella Condurache, Robin Bülow, Giovanni Donato Aquaro, Florian Andre, Le Thu-Thao, Dominika Suchá, Ahmed M. Salih, Roman Roy, et al.

    Introduction: Cardiovascular magnetic resonance (CMR) imaging offers precise quantification of cardiac structure and function. However, its clinical utility is often limited by the absence of robust, standardized reference ranges and severity grading thresholds. Aims: The aim of this study was to establish age-, sex-, and ethnicity-specific reference ranges and severity grading criteria for CMR-derived ventricular and atrial parameters in healthy adults, accounting for variations between two post-processing software tools. Methods and results: We analyzed CMR scans from the Healthy Hearts Consortium (HHC), which includes six multi-ethnic international cohorts. Images were automatically segmented using cvi42 (Circle Cardiovascular Imaging) and suiteHEART (Neosoft), with visual and statistical quality control. Ventricular and atrial volumes, myocardial mass, and ejection fractions were derived using short- and long-axis protocols; parameters were indexed to body surface area and height. We defined reference ranges as normal up to the 95% of the prediction interval (PI), and abnormalities as mild up to 99.73%, moderate at 99.73%, and severe at 99.99%, respectively. The final dataset included 4,624 women (51.0%) and 4,435 men (49.0%), with a mean age of 61 ± 13 years (range 18–83), and a multi-ethnic population (81.6% White, 5.6% South Asian, 5.3% Mixed/Other, 3.8% Black, 3.7% Chinese). Minor systematic differences were observed between cvi42 and suiteHEART, particularly in atrial parameters. Conclusions: Our work provides an evidence-based framework for CMR severity grading, offering age-, sex-, and ethnicity-stratified thresholds for mild, moderate, and severe deviations from the reference. These reference values support improved diagnostic accuracy, better risk stratification, and enhanced comparability of CMR findings worldwide. Graphical Footnote: This graphical abstract summarises the methodology and findings of our study on severity grading using cardiovascular magnetic resonance (CMR). It illustrates the dataset, quality control steps, software tools used and the derivation of population-specific reference ranges and severity grading classification. All reference ranges are available on the Healthy Hearts Consortium website ( www.healthy-hearts.org.uk ). Abbreviations: CMR: cardiovascular magnetic resonance; QC: quality control; EDV: end-diastolic volume; ESV: end-systolic volume; SV: stroke volume; EF: ejection fraction; LV: left ventricle.