Search Results (1,295)

Background and Objectives: Artificial intelligence (AI) has rapidly emerged as a promising tool for delivering personalized interventions in physical activity, exercise prescription, and weight management. AI technologies may facilitate individualized recommendations, behavioral support, and lifestyle modification through adaptive digital health solutions, although their effectiveness remains to be established across different populations and settings. However, the current evidence remains heterogeneous, and the practical implementation of AI in personalized training and weight management requires further evaluation. This scoping review aimed to summarize the current evidence regarding the application of artificial intelligence for the personalization of training and weight reduction, with particular emphasis on the types of AI technologies used, their reported outcomes, practical applications, and current limitations. Methods: A scoping review was conducted following a structured literature search of studies investigating AI-supported interventions related to physical activity, exercise, dietary behavior, and weight management. Eight studies involving diverse populations, intervention designs, and AI technologies were included. Data were extracted on study characteristics, AI technologies, intervention characteristics, reported outcomes, and research gaps. The literature search was subject to access-based restrictions, including the use of “Free full text” in PubMed and “Open Access” in Web of Science, as well as language restrictions. Results: The included studies investigated a wide range of AI technologies, including conversational chatbots, natural language processing systems, machine learning algorithms, computer vision applications, knowledge-based systems, and large language models. Selected studies reported favorable or modest changes in exercise adherence, physical activity participation, dietary behaviors, user engagement, and weight-related outcomes; however, the magnitude and consistency of these findings varied across studies. Personalized coaching, real-time feedback, and continuous behavioral support were common features of interventions reporting favorable outcomes. However, considerable heterogeneity existed across study designs, participant populations, intervention protocols, AI technologies, and outcome measures, and evidence regarding long-term effectiveness remains limited. The findings should be interpreted in the context of the adopted search strategy, including access-based restrictions and the inability to retrieve eight of 57 reports sought for retrieval, which may have contributed to availability bias. Conclusions: Current evidence suggests that artificial intelligence may have potential as a tool for supporting the personalization of training and weight management interventions, particularly through individualized behavioral support, feedback, and user engagement. However, the available evidence is heterogeneous and does not yet allow firm conclusions regarding the effectiveness or mechanisms of AI-supported interventions. AI should currently be viewed as a complement rather than a replacement for healthcare and exercise professionals. Future large-scale randomized controlled trials with longer follow-up periods and standardized outcome measures are needed to clarify the effectiveness, sustainability, and practical implementation of AI-supported interventions. Full article

Background: Patellar tendinopathy is a persistent load-related condition in athletes, and exercise is recommended as first-line care; however, the comparative value of specific tendon-loading prescriptions remains uncertain. This systematic review aimed to synthesize exercise-based interventions for adults with patellar tendinopathy, map reported loading prescription parameters, and estimate comparative effects when sufficiently comparable active-loading data were available. Methods: We conducted a PROSPERO-registered systematic review following PRISMA 2020. MEDLINE/PubMed, Web of Science Core Collection, Scopus, SPORTDiscus, and CENTRAL via Ovid/EBMR Reviews were searched. Two reviewers independently screened studies, extracted data, and assessed risk of bias. The primary outcome was VISA-P/VISA final score. Randomized active exercise/loading comparisons with extractable final-score data were pooled as mean differences on the original 0–100 scale using random-effects restricted maximum likelihood models with Hartung–Knapp adjustment. Results: The searches identified 942 records; 96 full-text reports were assessed and 40 reports were retained for qualitative synthesis. Six active exercise/loading comparisons contributed to the primary comparative synthesis. The pooled mean difference was 5.16 VISA-P/VISA points (95% CI 2.39 to 7.93) in favour of the first-listed intervention, with negligible statistical heterogeneity (tau2 = 0.000; I2 = 0.0%; Q = 3.50, p = 0.624). All contributing studies had some concerns in RoB 2, and GRADE certainty was low. Conclusions: Structured tendon-loading exercise remains central to rehabilitation. The pooled estimate should be interpreted as a small average comparative signal across clinically distinct active-loading contrasts, not as evidence that one specific loading strategy is universally superior. Full article

Background: Postpartum depression (PPD) affects about 15–20% of women worldwide. While observational studies have increasingly linked nature exposure to improved perinatal mental health, the extent to which nature-based approaches constitute a viable non-pharmacological intervention for PPD remains unclear. This scoping review and evidence mapping aimed to systematically characterize the current evidence base for nature-based approaches in PPD and perinatal populations. Methods: We searched PubMed, Web of Science, PsycINFO, Embase, Cochrane Library, Medline, CNKI, and WanFang Data from inception to December 2025. Two independent reviewers screened titles/abstracts and full texts against predefined inclusion criteria, with disagreements resolved through discussion. Empirical studies investigating any nature-based approach (NBA) (e.g., nature exposure, green space, forest therapy) in pregnant or postpartum women were incorporated. The review followed the PRISMA extension for Scoping Reviews (PRISMA-ScR) guidelines. No formal risk-of-bias assessment was conducted, consistent with scoping review methodology. The narrative synthesis and evidence mapping methodology were used to classify findings by population type and research design. Results: Of 1245 screened records, 11 studies fulfilled the inclusion criteria. To address the specific question of nature-based approaches for PPD, we classified evidence by its directness: studies on women with clinically significant PPD symptoms were treated as direct evidence; studies on general perinatal populations were treated as indirect evidence. Evidence mapping uncovered a significant deficiency: there are no randomized controlled trials (RCTs) that have specifically assessed NBA for women with PPD. Direct evidence was confined to two qualitative studies elucidating favorable experiences of nature engagement among postpartum women facing mental health challenges. Indirect evidence from observational studies (n = 10) generally supported associations between exposure to residential green space—especially tree canopy cover—and reduced risk of PPD and psychological distress, although some studies reported null findings. Physical activity was identified as a partial mediator in several studies. Only one pilot RCT focused on general postpartum women, indicating feasibility but not achieving significant depression reduction. Conclusions: Although encouraging observational data suggest a correlation between nature exposure and improved perinatal mental health, robust RCT evidence for structured Nature-Based Interventions in postpartum depression populations is lacking. Subsequent research must formulate and evaluate standardized, theoretically informed NBA protocols specifically tailored for women experiencing PPD. Full article

Background/Objectives: Dietary nitrate, long framed through food-safety concerns about N-nitroso compound formation, is now also recognised as a substrate of the nitrate–nitrite–nitric oxide pathway. This review aims to define the mechanistic, dietary and host conditions under which nitrate bioactivation becomes functionally relevant, with particular attention to its microbial determinants and to the level of inference the evidence actually supports. Methods: We conducted a narrative review supported by a structured literature search (PubMed, Scopus and Web of Science; 1 January 1976 to 14 February 2026; full-text, peer-reviewed, English-language, human-relevant sources; 148 sources retained, of which 93 contributed to the evidence synthesis), with narrative synthesis of mechanistic, interventional, observational and regulatory sources addressing dietary source and food matrix, enterosalivary metabolism, oral and gut microbial function, and health-related outcomes. A PRISMA-style flow diagram summarises the documented screening and inclusion process, and the complete database-specific search strategies are provided in ^{Supplementary Table S1}; no meta-analysis was performed because of substantial heterogeneity in designs and outcomes. Results: Within the canonical enterosalivary pathway, nitrate-to-nitrite bioactivation is predominantly microbiota-dependent and downstream conversion is chemically conditional: within the enterosalivary cycle, nitrate-reducing bacteria on the tongue dorsum generate the nitrite required for downstream nitric oxide formation, and its conversion in the stomach depends on pH and on matrix constituents. Dietary source and food matrix therefore govern both the delivered dose and the chemistry that follows, so vegetables, beetroot products, inorganic salts, drinking water and processed meat are not interchangeable exposure models. The oral microbiota is the principal microbial determinant of the response, whereas the gut microbiota acts as a context-dependent modifier of intestinal redox tone, barrier function and microbial ecology, supported by markedly weaker human evidence. Nitrate-rich sources reproducibly raise nitrate and nitrite biomarkers, with variable effects on blood pressure, vascular function and exercise efficiency, limited or inconsistent effects on cognition, cerebral blood flow and metabolic endpoints, and a safety profile whose interpretation depends on food matrix, dose, exposure pattern and host context rather than concentration alone. Conclusions: We propose the Source–Matrix–Microbiota–Host (SMMH) framework, in which biological impact depends on the interaction between dietary source and dose, food matrix, microbial nitrate-reducing capacity and host susceptibility, rather than on nitrate dose alone, and in which pathway-level, physiological and clinical evidence are kept explicitly distinct. The evidence base is mechanistically robust for the oral microbiota, considerably less defined for the gut microbiota, and variable at the level of validated clinical endpoints; it does not yet support source-independent guidelines or population-level recommendations. Full article

Financial return prediction increasingly relies on both financial text and structured market covariates, but adapting large language models across financial institutions remains difficult because raw data cannot be centralized and clients often exhibit heterogeneous, non-stationary market signals. This paper studies data-local federated financial return prediction with a frozen LLM, aiming to share useful cross-client adaptation while preserving client-specific predictive behavior. We propose a federated routed-adapter framework in which the server maintains a pool of lightweight adapter prototypes, each client selects a personalized mixture of these prototypes through projected directional routing, and local residual adapters are learned on private client data around the selected mixture. The server then maps uploaded residual updates back to the shared prototype space through an exact least-norm decomposition for communication-efficient aggregation. The framework keeps raw financial data and client-private prediction heads local, while uploaded residuals remain model updates and should not be interpreted as a formal privacy guarantee without additional mechanisms such as secure aggregation or differential privacy. Across FNSPID, Qlib CSI300/CSI800, and Open FinLLM forecasting benchmarks, our method achieves the best overall performance, improving CSI300 RankIC from $0.082$ to $0.087$ over the strongest federated PEFT baseline and reducing FNSPID MAE from $0.00537$ to $0.00482$. These results suggest that compositional shared adaptation with local residual personalization is a practical direction for financial LLM deployment under data-local, communication-constrained, and heterogeneous federated settings. Full article

Urban Digital Twins excel at modeling physical infrastructure but remain structurally limited in capturing the qualitative, experiential dimensions of urban life—particularly sense of place, which empirical research links to civic stewardship and long-term sustainability. This study investigates whether generative AI can serve as a participatory elicitation interface for surfacing these missing human data layers. Through a mixed-methods experimental design, 24 residents of Austin, Texas, each selected a personally meaningful public urban space and created visual representations using both hand-drawn sketching and iterative co-creation with the text-to-image model DALL-E. Pre- and post-experiment surveys and semi-structured interviews captured participants’ perceptions of the outputs and self-reported shifts in place awareness. The findings reveal a dialectical tension: DALL-E consistently defaulted to generic visual archetypes, overriding participants’ localized descriptions. However, this algorithmic homogenization paradoxically deepened participants’ sense of place through a process we term ‘validation by contrast’—residents utilized the AI’s inaccurate outputs as a foil to consciously articulate what made their environments authentically meaningful. These findings suggest that for human-centric Digital Twins, the actionable data lies not in the AI-generated image itself, but in the negotiation process through which residents defend and crystallize their authentic spatial identity. Full empirical validation of this pattern, including systematic comparison across representation modalities, is reserved for future work. Full article

Metallurgical R&D search is difficult for a practical reason: useful evidence is rarely defined by one keyword. A production-support question can depend at the same time on material grade, process route, defect mechanism, property, test method, and numerical conditions. Conventional retrieval-augmented generation (RAG) pipelines largely treat document chunks as independent text and can therefore miss relations that matter for process monitoring, fault diagnosis, and engineering decision support. We evaluate a confidence-adaptive graph-enhanced retrieval layer for metallurgical RAG in a controlled synthetic benchmark with explicitly specified generation and evaluation rules. The benchmark contains 300 generated heterogeneous records derived from a seven-block source distribution and 30 material–process–defect–property archetypes, together with 180 frozen queries: 60 exact, 60 paraphrased, and 60 multi-hop. The main run evaluates robustness to incomplete structured metadata, with 10% missing and 4% erroneous categorical fields. Entity and relation extraction from raw documents is outside the evaluated scope. We compare BM25, TF-IDF, latent semantic analysis, a sparse + dense hybrid, graph-only retrieval, two ablations, and the proposed adaptive hybrid. On the complete query set, the proposed method obtains MRR = 0.992, Precision@5 = 0.980, Recall@10 = 0.859, and nDCG@10 = 0.948. Relative to the sparse + dense hybrid, nDCG@10 increases by 0.186 (24.4%); the paired 95% bootstrap interval is in the range of 0.166–0.206, and the Holm-adjusted Wilcoxon p-value is 2.59 × 10⁻²⁹. Under severe degradation with 40% missing and 16% erroneous metadata, the adaptive method retains mean nDCG@10 = 0.791, compared with 0.650 for graph-only retrieval and 0.762 for the metadata-independent sparse + dense hybrid. A 5000-run Monte Carlo analysis estimates 6763 chunks and 58.70 MB for indexed vectors plus metadata at a 512-token chunk size and 64-token overlap. The results show how the retrieval rule behaves under controlled conditions; they are not evidence of plant-level effectiveness or of the quality of generated answers. Those questions require external, expert-labeled validation. Full article

Underwater visual degradation makes it difficult for the image modality to represent video semantics, while label sparsity in underwater scenes leads to weak inter-category correlations, thereby degrading the performance of multi-label classification. To address these issues, this paper proposes a Semantic-Enhanced Underwater Videos Multi-Label Classification Network Based on Structural Graph Convolution (SEMGCN). Specifically, the proposed method first disentangles the image and text modalities into shared and private representations, and enhances feature representation capability through orthogonal constraints and feature reconstruction. Moreover, a Cross-Modal Category-Aware Module (CCAM) is constructed to model interactions between image and text features and perform bidirectional cross-attention with category-label text embeddings, thereby generating category-aware initial node representations. Furthermore, to alleviate the limitation of semantic propagation caused by sparse label co-occurrence, a Structural Graph Convolutional Network (SGCN) is proposed. By integrating explicit co-occurrence relationships with implicit structural similarity relationships, the proposed model collaboratively captures both explicit and latent semantic associations, thereby improving multi-label classification performance under label-sparse conditions. Experiments were conducted on the self-constructed Underwater Video Multi-label Classification Dataset (UVMC) and the public MLSV2018 dataset. The experimental results show that SEMGCN achieves Average Precision scores of 0.8645 and 0.8388 on UVMC and MLSV2018, respectively, demonstrating the effectiveness of the proposed method. Full article

Fault risks in marine diesel engines (MDEs) propagate across coupled subsystems and evolve with component degradation, but existing methods rarely integrate accident narratives, causal structure, and time-varying reliability. This study develops a novel framework that integrates large language models (LLMs), rough–fuzzy DEMATEL, interpretive structural modeling (ISM), and Bayesian networks (BNs) with service-time-dependent priors for time-varying failure analysis. First, the risk-influencing factors (RIFs) are extracted from accident and maintenance records using LLMs, text embeddings, semantic clustering, and expert consolidation. Rough–fuzzy DEMATEL and ISM are used to identify causal relationships and the hierarchical structure. The RIFs, bottom-level components, and target failure are then mapped into a multilayer BN parameterized using Noisy-OR relationships and Weibull-derived time-varying priors. In a case study of MDE hard starting, 338 cause descriptions from 35 records yielded 12 RIFs, with semantic coverage above 93% across four evaluation models. When hard starting was observed, the posterior probability of mechanical failure of the fuel injection system reached 60.92%, compared with 36.91% for governor and mechanical actuation system failure. Over 0–10,000 h of cumulative service, the model-inferred probability of hard starting during a single starting attempt increased from 38.09% to 75.23% under the specified model parameterization. The framework supports causal interpretation, troubleshooting prioritization, and service-time-dependent maintenance prioritization. Full article

Background: Preeclampsia remains a major cause of maternal and perinatal morbidity and mortality, and machine learning (ML) and artificial intelligence (AI) models have increasingly been proposed for risk prediction, diagnosis, monitoring, and prognosis. However, the extent to which key methodological details are visible from abstracts and bibliographic metadata remains unclear. Methods: We conducted an abstract-level systematic mapping review of ML/AI studies for preeclampsia (PE), hypertensive disorders of pregnancy (HDP), and directly related complications. Records were identified from ACM Digital Library, Elsevier/ScienceDirect, IEEE Xplore, PubMed, Scopus, SpringerLink, and snowballing. After DOI-based deduplication, abstract-availability checks, technical metadata filtering, and a two-stage eligibility audit, 29 primary ML/AI studies were retained for the final synthesis. Two authors independently re-audited the 145 technically retained records for eligibility. Initial agreement was 131/145 records (90.3%); for the binary include/exclude decision, Cohen’s kappa was approximately 0.764, indicating substantial agreement. Extraction was limited to title, abstract, year, DOI, and bibliographic metadata using a predefined RQ-aligned schema. Therefore, all frequencies represent information explicitly extractable at the abstract level and should be interpreted as lower-bound reporting estimates rather than full-text prevalence estimates. Results: Among the 29 eligible studies, tree-based models were the most frequently identifiable family (17/29), followed by linear/generalized linear models (15/29), support vector machines (9/29), and neural networks (8/29). The most frequently identifiable input modalities were EHR/clinical variables (21/29), demographics (10/29), laboratory variables (7/29), vital signs/blood pressure (7/29), and ultrasound/Doppler information (6/29). AUC was mentioned in 20/29 abstracts, but a numeric AUC value was extractable into the structured field in 14/29. These values were summarized descriptively and were not pooled because outcomes, prediction horizons, populations, predictors, and validation strategies were heterogeneous. External validation was explicitly identifiable in 8/29 abstracts, explainability-related methods in 5/29, and repository or dataset-source information in 1/29. Conclusions: The abstract-level evidence shows an active but heterogeneous ML/AI literature for PE/HDP prediction and related outcomes. However, abstracts often provide insufficient detail for robust comparative assessment of model validity, calibration, clinical readiness, reproducibility, and implementation. Future work should prioritize full-text evidence synthesis, standardized reporting, external and temporal validation, calibration, equity assessment, clinically meaningful explainability, and reusable data/model resources. Full article

Named entity recognition (NER) is a key technique for extracting entities such as equipment, components and defect types from GIS defect texts, providing a basis for subsequent knowledge graph construction. However, GIS defect texts contain many engineering structures, including engineering abbreviations, equipment numbers and phase identifiers, making it difficult for general-purpose models to stably recognize their semantic associations and entity boundaries. To address this problem, this paper proposes an engineering format-aware masking strategy. The strategy identifies candidate fragments using format rules for phase identifiers, measurement value-unit patterns and engineering abbreviations and preferentially selects them as perturbation targets to strengthen the model’s understanding of engineering structures and their contextual relationships. Bidirectional long short-term memory is used to extract bidirectional sequence features, and a conditional random field is used to model transition constraints between labels and obtain the globally optimal label sequence. The results show that the proposed model achieves precision, recall and F₁ scores of 0.89, 0.92 and 0.90, respectively. The analysis indicates that the proposed method improves entity recognition for phase-related structures, engineering abbreviations and equipment hierarchy fragments. Full article

Background: Mobile health interventions offer a potential solution to adherence challenges, yet evidence regarding their collective efficacy in schizophrenia spectrum disorders has not been formally synthesized. This systematic review evaluates the impact of mobile health (mHealth) interventions on medication adherence as a primary outcome and on daily functioning and psychotic symptoms as secondary outcomes in individuals with schizophrenia spectrum disorders. Methods: Using the Population, Intervention, Comparison, Outcome (PICO) framework, a systematic search was conducted across multiple databases to identify relevant randomized controlled trials (RCTs) evaluating mHealth strategies for medication adherence in adults with schizophrenia spectrum disorders. The PubMed, CINAHL, PsycINFO, EMBASE, and JBI databases were searched from inception until 24 February 2026, using combinations of search terms such as “Schizo” OR “Psychos” AND “mHealth” OR “Digital Health” AND “Medication Adherence”. Data extraction was conducted using a standardized data extraction table and narratively synthesized. This review adheres to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, ensuring structured and comprehensive reporting of the findings. Results: Fourteen randomized controlled trials (RCTs) with 1717 participants were included in this review. Four studies evaluated text messaging interventions, four employed phone call interventions, three used electronic medication monitoring systems, and three used mobile applications. Nine of the fourteen studies reported statistically significant improvements in medication adherence. For secondary outcomes, the results were highly inconsistent: only three studies demonstrated significant reductions in psychotic symptoms, and none showed benefits for daily functioning. While various theoretical frameworks, such as the Health Belief Model and Cognitive Behavioral Therapy and intervention modalities, were utilized, the overall evidence was limited by high clinical heterogeneity and a lack of robust long-term data. Conclusions: mHealth interventions, particularly text messaging and mobile applications, demonstrate clear potential to improve medication adherence in individuals with schizophrenia spectrum disorders. Given the high heterogeneity and lack of long-term evidence, future research should prioritize standardized outcome measurements, rigorous designs, and extended follow-up periods to confirm clinical utility. Full article

Background/Objectives: Clinical alarm management in continuous patient monitoring systems remains a major challenge in intensive care units (ICUs). The high frequency of nonactionable alarms contributes to alarm fatigue among healthcare professionals, potentially compromising patient safety, clinical surveillance, and the quality of nursing care. As nurses play a central role in monitoring critically ill patients, evidence-based alarm management strategies are essential to improve the quality of care and patient outcomes. This review aimed to synthesize the scientific evidence on nursing interventions for managing clinical alarms in continuous monitoring systems and their contribution to patient safety and quality of care among critically ill patients. Methods: An integrative literature review was conducted following the methodological framework proposed by Whittemore and Knafl. A comprehensive literature search was performed in CINAHL Plus with Full Text, MEDLINE with Full Text, Supplemental Index, Complementary Index, Academic Search Index, the Directory of Open Access Journals (via EBSCOhost^®), PubMed, Scopus, and ScienceDirect. Studies published between 2020 and 2025 in English, Portuguese, or Spanish were eligible for inclusion. Methodological quality was assessed using the Joanna Briggs Institute critical appraisal tools. Nine studies met the eligibility criteria and were included in the final synthesis. Results: The identified interventions were grouped into five main domains: alarm parameter customization, nurse education and training, implementation of structured protocols and alarm management bundles, optimization of monitoring devices and sensors, and interventions addressing human and organizational factors. The evidence indicated that multifaceted and integrated interventions were associated with reductions in nonactionable alarms and improvements in clinical responses to alarms. However, their effectiveness in reducing alarm fatigue appeared to depend on organizational culture, staff engagement, and contextual factors. Conclusions: Effective alarm management requires the consistent implementation of evidence-based nursing interventions. Nurses play a pivotal role in promoting patient safety through clinical expertise, technological competence, and adherence to structured alarm management practices. Strengthening alarm management strategies may contribute to improved quality of care, safer care delivery, and better outcomes for critically ill patients. Full article

Engineering schematics, such as electrical, mechanical, piping and instrumentation diagrams, record how industrial plants are built and operated, but most of them survive only as images or scanned sheets that software cannot read. Digitizing them by hand is slow and error-prone: an expert must find and classify hundreds of symbols, read dense technical text, and work out which label belongs to which component. Progress with learning-based methods has been held back on two fronts at once. There are almost no annotations that connect a text label to its symbol, and the drawings themselves are usually confidential, so even unlabeled sheets rarely reach the public domain. We address this with a system that turns a drawing into a structured, queryable graph: it detects and classifies the graphical components with an object detector, recovers the technical text, and then resolves which label belongs to which component. Our contributions are threefold: (i) the first at-scale dataset of manually annotated text-to-symbol links for industrial schematics; (ii) a complete, deployable digitization system combining tiled detection with sliced inference, off-the-shelf OCR, and a text-to-symbol association stage; and (iii) a rigorous, leakage-free benchmark of association methods. Under an observable-only candidate protocol, we find that on logic circuits association is dominated by geometry: a simple pairwise model reaches about 99% top-1 and a graph neural network matches but does not exceed it, whereas the denser P&IDs still benefit from a geometric rule-based chain. Detection reaches an mAP@50 of 0.995 on logic circuits and about 0.91 across the 107-class P&ID taxonomy. The system produces a partial semantic graph; connecting lines and flow direction are not extracted. Full article

Background/Objectives: Haematopoietic stem cell transplantation (HSCT) causes substantial morbidity in children and young people (CYP). Although prehabilitation benefits adults with cancer, its role in paediatric HSCT remains underexplored. This study aimed to describe current pre-transplant assessment and supportive care practices across paediatric HSCT centres, explore healthcare professional perspectives on prehabilitation implementation, and integrate patient and public involvement (PPI) to inform future intervention design. Methods: A cross-sectional survey was administered to healthcare professionals across 16 paediatric HSCT centres in the UK and the Republic of Ireland, alongside semi-structured PPI discussions conducted with eight children, young people, and caregivers with direct experience of paediatric HSCT. Survey data were analysed descriptively, free-text responses by qualitative content analysis, and PPI by reflexive thematic analysis. Results: Forty-nine eligible responses were received from healthcare professionals. Formal prehabilitation services were reported by 27 (55.1%) respondents, with substantial within-centre variation. Nutritional advice was the most consistently delivered component, 36 (72%), whilst physical activity interventions were the least consistently provided, 13 (26%). Workforce limitations were identified as the dominant barrier by 37 (95%) respondents. PPI findings described provision as reactive and inconsistent, with families expressing preference for flexible, hybrid, and family-centred delivery models. Conclusions: Prehabilitation provision in paediatric HSCT is variable, often informal, and limited by workforce capacity. CYP and caregivers expressed a desire for prehabilitation, particularly through flexible, hybrid (face-to-face and online), and family-centred approaches. Future research should prioritise co-design and feasibility testing of safe, individually tailored programmes that can be integrated into paediatric HSCT pathways. Full article