Federated learning (FL) enables privacy-preserving model training across distributed Electronic Health Records (EHRs), but its deployment remains limited by data-view heterogeneity, where institutions maintain incompatible local schemas. Most existing methods address this by enforcing flat, aligned data views, which require extensive cross-site preprocessing and manual harmonisation that often discards client-specific features, or by projecting inputs into a shared latent space, which sacrifices interpretability. We propose a modelling shift from conventional FL with vectorised inputs to a symbolic, relation-centric framework, where each client organises its EHR data as a structured, type-aware relational graph. This enables client-specific inference without requiring schema alignment and supports FL across heterogeneous data views. To model over these symbolic structures, we introduce an architecture that combines relation-aware message passing with a learnable feature relevance mechanism, jointly enabling accurate local predictions and client-specific interpretability while supporting parameter sharing across clients. Beyond strong performance on three real-world EHR datasets exhibiting data-view heterogeneity, we further show that our framework supports multimodal FL under modality-level heterogeneity. Using MC-MED, a publicly available multimodal emergency department dataset, we demonstrate that our method accommodates clients with partially missing modalities, highlighting its robustness and scalability in real-world clinical settings.

Federated Learning (FL) enables collaborative clinical modelling across distributed electronic health records (EHRs) without sharing sensitive patient data. However, variations in medical practice, documentation standards, and data collection across institutions create data-view heterogeneity, where clients possess different or only partially overlapping clinical feature sets. This misalignment hinders the use of standard FL methods. Existing approaches rely on complex preprocessing and manual harmonisation, which can cause information loss, reduce data utility, limit scalability, and restrict client-specific personalisation. To address these limitations, we propose Personalised Attention-based Federated Graph Network (PAFNet), a scalable FL framework that enables meaningful parameter exchange across heterogeneous clients by mapping their distinct data-views into a shared latent space through client-specific projection layers. It then applies a personalised adaptation mechanism using trainable parameter masks, allowing each client to selectively incorporate global model parameters relevant to its own feature set. This design preserves local specificity, improves generalisation, and removes the need for heavy manual preprocessing common in existing approaches. Across CURIAL, eICU, and MIMIC-III datasets, PAFNet consistently outperformed state-of-the-art data-view heterogeneity FL baselines, demonstrating strong generalisation under substantial differences in client feature sets. By enabling effective personalisation and cross-institutional knowledge sharing without extensive harmonisation, PAFNet offers a robust and scalable solution for the federated training of clinical models in data-view heterogeneous environments.

This paper presents Adaptive Parameter Optimisation (APO), a novel framework for optimising shared models across multiple clinical tasks, addressing the challenges of balancing strict parameter sharing—often leading to task conflicts—and soft parameter sharing, which may limit effective cross-task information exchange. The proposed APO framework leverages insights from the lazy behaviour observed in over-parameterised neural networks, where only a small subset of parameters undergo any substantial updates during training. APO dynamically identifies and updates task-specific parameters while treating parameters associated with other tasks as protected, limiting their modification to prevent interference. The remaining unassigned parameters remain unchanged, embodying the lazy training phenomenon. This dynamic management of task-specific, protected, and unclaimed parameters across tasks enables effective information sharing, preserves task-specific adaptability, and mitigates gradient conflicts without enforcing a uniform representation. Experimental results across diverse healthcare datasets demonstrate that APO surpasses traditional information-sharing approaches, such as multi-task learning and model-agnostic meta-learning, in improving task performance.

Federated Learning (FL) involves a server aggregating local models from clients to compute a global model. However, this process can struggle to position the global model in low-loss regions of the parameter space for all clients, resulting in subpar convergence and inequitable performance across clients. This issue is particularly pronounced in non-IID settings, common in clinical contexts, where variations in data distribution, class imbalance, and training sample sizes result in client heterogeneity. To address this issue, we propose a mode connectivity-based FL framework that ensures the global model resides within the overlapping low-loss regions of all clients in the parameter space. This framework models the low-loss regions as non-linear mode connections between the current global and local models, and optimises to identify an intersection among these mode connections to define the new global model. This approach enhances training stability and convergence, yielding better and more equitable performance compared to standard FL frameworks like federated averaging. Empirical evaluations across multiple healthcare datasets demonstrate the benefits of the proposed framework.

Online mental health interventions are increasingly important in providing access to, and supporting the effectiveness of, mental health treatment. While these technologies are effective, user attrition and early disengagement are key challenges. Evidence suggests that integrating a human supporter into such services mitigates these challenges, however, it remains under-studied how supporter involvement benefits client outcomes, and how to maximize such effects. We present our analysis of 234,735 supporter messages to discover how different support strategies correlate with clinical outcomes. We describe our machine learning methods for: (i) clustering supporters based on client outcomes; (ii) extracting and analyzing linguistic features from supporter messages; and (iii) identifying context-specific patterns of support. Our findings indicate that concrete, positive and supportive feedback from supporters that reference social behaviors are strongly associated with better outcomes; and show how their importance varies dependent on different client situations. We discuss design implications for personalized support and supporter interfaces.

High prevalence of mental illness and the need for effective mental health care, combined with recent advances in AI, has led to an increase in explorations of how the field of machine learning (ML) can assist in the detection, diagnosis and treatment of mental health problems. ML techniques can potentially offer new routes for learning patterns of human behavior; identifying mental health symptoms and risk factors; developing predictions about disease progression; and personalizing and optimizing therapies. Despite the potential opportunities for using ML within mental health, this is an emerging research area, and the development of effective ML-enabled applications that are implementable in practice is bound up with an array of complex, interwoven challenges. Aiming to guide future research and identify new directions for advancing development in this important domain, this article presents an introduction to, and a systematic review of, current ML work regarding psycho-socially based mental health conditions from the computing and HCI literature. A quantitative synthesis and qualitative narrative review of 54 papers that were included in the analysis surfaced common trends, gaps, and challenges in this space. Discussing our findings, we (i) reflect on the current state-of-the-art of ML work for mental health, (ii) provide concrete suggestions for a stronger integration of human-centered and multi-disciplinary approaches in research and development, and (iii) invite more consideration of the potentially far-reaching personal, social, and ethical implications that ML models and interventions can have, if they are to find widespread, successful adoption in real-world mental health contexts.

Wheezing is common among children and ~50% of those under 6 years of age are thought to experience at least one episode of wheeze. However, due to the heterogeneity of symptoms there are difficulties in treating and diagnosing these children. `Phenotype specific therapy' is one possible avenue of treatment, whereby we use significant pathology and physiology to identify and treat pre-schoolers with wheeze. By performing feature selection algorithms and predictive modelling techniques, this study will attempt to determine if it is possible to robustly distinguish patient diagnostic categories among pre-school children. Univariate feature analysis identified more objective variables and recursive feature elimination a larger number of subjective variables as important in distinguishing between patient categories. Predicative modelling saw a drop in performance when subjective variables were removed from analysis, indicating that these variables are important in distinguishing wheeze classes. We achieved 90%+ performance in AUC, sensitivity, specificity, and accuracy, and 80%+ in kappa statistic, in distinguishing ill from healthy patients. Developed in a synergistic statistical - machine learning approach, our methodologies propose also a novel ROC Cross Evaluation method for model post-processing and evaluation. Our predictive modelling's stability was assessed in computationally intensive Monte Carlo simulations.