192815 research outputs found
Sort by
Protocol for the PROSECCA study: a new approach for predicting radiotherapy outcome using artificial intelligence and electronic population-based healthcare data
No full textIntroduction: Within the UK there are 33 deaths every day from prostate cancer, second only to lung cancer as the most common cause of cancer death in males in the UK. Of the 55 000 new cases each year, up to 50% of these patients will receive radiotherapy either alone or after prostatectomy. Although there have been significant improvements in the accuracy of radiotherapy delivery leading to better tumour targeting and a reduction in dose to normal tissues, significant permanent genito-urinary or gastrointestinal-related side effects are all too common. With nearly 80% of patients with prostate cancer surviving for 10 years or more, minimising life-limiting radiation damage to normal tissues is vitally important. However, at present, it is not possible to identify which patients will suffer a poorer outcome after radiotherapy. The aim of this study, improving radiotherapy in PROState cancer using EleCtronic population-based healthCAre data (PROSECCA), is to do this by using the existing information in a patient’s digital healthcare record. By linking primary, secondary and tertiary clinical data, including digital image information, with radiotherapy treatment plans and outcome data, the PROSECCA study will identify de novo predictive biomarkers of radiation response and provide clinicians with a tool to individualise a radiotherapy dose and plan to maximise cure and minimise toxicity.
Methods and analysis: The PROSECCA study is a large multidisciplinary project, the purpose of which is to analyse healthcare records from up to 15 000 patients with prostate cancer who underwent radiotherapy in the treatment of their cancer in Scotland between 2010 and 2022. Through the linkage of data obtained specifically for radiotherapy and data held within each patient’s unique electronic health record (EHR), the factors that indicate why some patients have a poor response to treatment, or an increased risk of side effects from radiation, will be identified. This will be made possible by the use of artificial intelligence and machine learning (AL/ML), which will help to identify at-risk patients earlier and allow adaptation of their treatment accordingly.
Ethics and dissemination: The study is being conducted in accordance with the ethical principles set out in the Declaration of Helsinki and Good Clinical Practice that respects and protects the rights, and maintains confidentiality, of all trial participants. The study protocol (V.1.0) was reviewed by the South Central Oxford A Research Ethics Committee (REC) on 13 December 2021 and received a favourable opinion subject to each National Health Service (NHS) organisation confirming permission for patients treated within their area. Approval for the use of unconsented healthcare record data for patients included in the study and treated at one of the five Scottish Cancer Centres required an application to the NHS Scotland Public Benefit and Privacy Panel for Health and Social Care (HSC-PBPP). Full approval from the HSC-PBPP panel was received on 1 July 2024, which covered the use of pseudoanonymised EHR data for all patients participating in the study. The study is publicly listed on the NHS Health Research Authority site, with IRAS ID 306245 and REC reference 21/SC/0402. Dissemination of the study findings will take place through field-leading cancer, radiation oncology and medical physics journals. All manuscripts will be approved by the main study team and authorship determined by mutual agreement
Dimos: Diffusion model with unified sequential state space for session-based recommendation
No full textSession-based recommendation aims to predict the next item based on the user–item interactions within the current session. Many existing methods adopt discriminative approaches to learn specific preference representations, while few methods introduce generative approaches to learn underlying preference distributions, failing to handle limited and noisy interactions effectively. Moreover, naive implementations of generative models face a trade-off between effectiveness and efficiency, limiting their practical utility. To address these challenges, we propose Dimos, a dual-branch framework comprising an exploring branch and an exploiting branch, which leverage diffusion models and attention networks to capture implicit and explicit preferences, respectively. At the core of Dimos is Bi-MaKAN, a novel backbone architecture featuring a pair of parameter-sharing bidirectional Mamba blocks and a Kolmogorov–Arnold network-based feature fusion layer, designed to enhance both performance and efficiency. To further improve generalization and reduce overfitting, we unify the sequential state spaces of both branches. Additionally, we introduce a linearly weighted fusion mechanism that integrates preference representations from both branches, enabling flexible adjustment of implicit and explicit preference contributions during training and inference. Extensive experiments on three real-world benchmark datasets demonstrate the superiority of Dimos, achieving up to 2.79% improvement in Recall, 3.09% in Mean Reciprocal Rank (MRR), and 3.00% in Normalized Discounted Cumulative Gain (NDCG) over state-of-the-art baselines. Efficiency evaluations show substantial gains, with reductions of 94.32% in Graphics Processing Unit (GPU) memory usage, 66.81% in training time, and 98.80% in inference time. In-depth analyses reveal a collaborative effect between the two branches during both training and inference, with dataset scale modulating their relative importance
Screening of purine nucleoside analogues against intracellular Toxoplasma gondii
No full textToxoplasmosis remains a world-wide public health concern, especially for the immunocompromised. Although this population segment is increasing due to therapeutic interventions, organ transplants and infections including HIV, treatment relies almost exclusively on sulfadoxine and pyrimethamine, antifolates developed against malaria but with only moderate efficacy against acute toxoplasmosis and no effect on the chronic stage. Here we explore whether 7-substituted analogues of 7-deazaadenosine (tubercidin) that have shown remarkable efficacy against other protozoan pathogens, might also show anti-toxoplasmic activity. Tubercidin and a series of eleven 7-substituted analogues including 2’-deoxy and 3’-deoxyribofuranoses was tested against intracellular Toxoplasma gondii tachyzoites. The test compounds yielded EC 50 values between 0.012 and 1.72 µM, well below those of the control drug sulfadiazine (11.9 µM) and the previously identified purine analogue adenosine arabinoside (Ara-A; 11.4 µM). The tubercidin analogues displayed at most moderate toxicity to HFF cells, with the most efficacious compound, 7-(3,4-di-Cl-phenyl)-3’-deoxytubercidin (FH8513) reaching a selectivity index of >2500. These nucleosides are most likely taken up by T. gondii through one of the four Equilibrative Nucleoside Transporters (ENTs) encoded by the parasites. However, deletion of TgENT2 and/or TgENT3 had no effect on the EC 50 values, and deletion of TgAT1 actually sensitised the tachyzoites to most of the tubercidin analogues. We propose that these nucleosides are internalised through the TgENT1 uridine transporter and that the sensitisation in ΔTgAT1 cells is the result of reduced uptake of adenosine that competes with the tubercidin analogues for metabolic enzymes such as adenosine kinase
Masked Generative Policy for Robotic Control
No full textWe present Masked Generative Policy (MGP), a novel framework for visuomotor imitation learning. We represent actions as discrete tokens, and train a conditional masked transformer that generates tokens in parallel and then rapidly refines only low-confidence tokens. We further propose two new sampling paradigms: MGP-Short, which performs parallel masked generation with score-based refinement for Markovian tasks, and MGP-Long, which predicts full trajectories in a single pass and dynamically refines low-confidence action tokens based on new observations. With globally coherent prediction and robust adaptive execution capabilities, MGP-Long enables reliable control on complex and non-Markovian tasks that prior methods struggle with. Extensive evaluations on 150 robotic manipulation tasks spanning the Meta-World and LIBERO benchmarks show that MGP achieves both rapid inference and superior success rates compared to state-of-the-art diffusion and autoregressive policies. Specifically, MGP increases the average success rate by 9% across 150 tasks while cutting per-sequence inference time by up to 35×. It further improves the average success rate by 60% in dynamic and missing-observation environments, and solves two non-Markovian scenarios where other state-of-the-art methods fail. Further results and videos are available at: https://anonymous.4open.science/r/masked_generative_policy-8BC6
Sustained use of compensatory technology following brain injury: experiences from Neumind users
Get PDFassistive technology for cognition (atc), particularly smartphone-based apps, can significantly enhance rehabilitation outcomes for individuals with acquired brain injury (aBi). Despite demonstrated efficacy, challenges persist in initial uptake and long-term engagement with such technologies. this study employed a mixed-methods approach to examine factors influencing sustained engagement with the Neumind app over 13–14 months among individuals with aBi. twelve adult participants were interviewed, and their app usage data was quantitatively analysed over nine months post-interview. eleven participants used the app when interviewed 4–5 months after download (3 highly active, 3 moderately active and 5 occasional users). the six active users maintained their use nine months later. Qualitative thematic analysis identified six major themes impacting sustained atc engagement: introduction and initial Use (recommendation by others, ease of setup), Perceived Benefits (increased confidence, independence), technical and Functional challenges (technical issues, reliability, simplicity), User customisation Preferences (reminder features, calendar scheduling), community and social support (sense of belonging, helping others), and integration into Rehabilitation and Daily life (integration with rehabilitation, habit formation, established strategies, timing of introduction). Participants who maintained consistent use reported tangible benefits, effective habit formation, emotional engagement through community participation, and successful integration into rehabilitation routines. Non-users faced technical difficulties, perceived irrelevance of content, and preferred existing strategies. this study underscores the critical role of early introduction, ease of use, community integration, and habit formation in sustaining atc engagement. addressing usability barriers, enhancing customisation, and incorporating structured social support may significantly improve long-term use and effectiveness of atc interventions in neuro-rehabilitation
Identification of serum protein biomarkers for pre-cancerous lesions associated with pancreatic ductal adenocarcinoma
Get PDFPancreatic ductal adenocarcinoma (PDAC) has poor prognosis as early-stage asymptomaticity leads to late-stage diagnoses. Strategies to detect PDAC earlier or identify high-risk individuals are therefore paramount. Here, we report results from genetically engineered mice and PDAC patients that identify serum proteins associated with pancreatic intraepithelial neoplasms (PanINs), the most common PDAC precursor, and early-stage PDAC. Initially, we screened previously described PanIN-abundant mice, harbouring pancreatic and duodenal homeobox 1 (Pdx1)-Cre, Lox-STOP-Lox-KrasG12D/+ and floxed alleles of essential autophagy genes autophagy-related 7 (Atg7) or autophagy-related 5 (Atg5). Sera from these mice were assessed by proteomics and hits were compared to those in Lox-STOP-Lox-KrasG12D/+ Lox-STOP-Lox-Trp53R172H/+ Pdx1-Cre (KPC) mice, which closely recapitulate human disease, and early-stage (I–II) PDAC patients. Levels of inter-alpha-trypsin inhibitor heavy chain H3 (ITIH3) were significantly elevated in all three screens, with complement C5, complement factors B and H (CFB/CFH), and monocyte differentiation antigen CD14 increased in KPC mice and PDAC patients; and all were significantly increased co-ordinately in PDAC according to disease stage. Serum levels of C5, CFH and CD14 together constitute a novel panel for identifying PanINs and early-stage PDAC with confidence, and when combined with additional screening, could help increase survival from this dismal disease
Prospective evaluation of artificial intelligence integration into breast cancer screening in multiple workflow settings: The GEMINI study
Get PDFArtificial intelligence (AI) tools can improve breast screening performance but different screening sites have varying needs. Here the GEMINI prospective evaluation of 10,889 women, within one UK region, used both live AI integration and simulations to model 17 different ways AI could be used in breast screening. All women received routine care. One AI tool was assessed. When the AI tool recommended recall but routine double reading did not, cases underwent additional human review, detecting 11 additional cancers. The primary AI workflow could improve cancer detection by 10.4% (1 per 1,000), maintain the recall rate (0.8% reduction) and reduce workload by up to 31%. Other workflow variations significantly improved all measured metrics (superiority in cancer detection rate, recall rate, positive predictive value (PPV), sensitivity and specificity) with up to 36% workload savings. Different AI integrations in breast screening could offer various clinical and operational gains, allowing for adaptation to local healthcare needs
Early bone health assessment is indicated in Parkinson's: risk factor analysis in a UK population-based cohort study
Get PDFBackground:
Osteoporosis and major osteoporotic fracture are more common in PD than controls, but evaluation of sociodemographic factors and preventive treatment is limited.
Methods:
In a UK population-based nested case-control study, incidence rates of osteoporosis and major osteoporotic fracture were calculated, and risk factors and bone health treatments analyzed by multivariable regression.
Results:
Falls, osteoporosis and major osteoporotic fracture were more likely before a diagnosis of PD compared to controls (P < 0.0001). After diagnosis, incident osteoporosis, HR 1.96 (1.90, 2.02) and incident fractures, HR 2.16 (2.11, 2.21) were more likely in PD than controls, both P < 0.005. Risks of fracture were increased in PD cases during 3 consecutive time periods post-diagnosis: 0–7 years, HR 2.15 (2.09, 2.21), 7–14 years, HR 2.36 (2.24, 2.48), and 14–21 years, HR 1.88 (1.62, 2.17), all P < 0.0001. Risks of osteoporosis and fracture increased with older age, female sex, greater deprivation and White ethnicity. In PD, anti-osteoporosis treatment was underutilized in men, OR 0.25 (0.24, 0.26) versus women, being lower than the relative risks of osteoporosis in men, HR 0.35 (0.32, 0.37) and fracture in men, HR 0.60 (0.57, 0.63). There was also underutilization of anti-osteoporosis treatment in the most deprived quintile, HR 0.85 (0.80, 0.90), despite significantly higher rates of osteoporosis, HR 1.14 (1.01, 1.28) and fracture, HR 1.17 (1.06, 1.30).
Conclusion:
The increased bone health risks at all stages including the prodrome, and across multiple risk categories, emphasizes the need for early bone health assessment and increased anti-osteoporosis treatment rates
Isorhythmic atrioventricular dissociation in an alert normothermic cat
Get PDFAn 11-year-old male neutered Birman cat was referred for investigation of pyelonephritis. On physical examination gallop sounds and bradycardia were noted. There was mild azotemia and marked neutrophilia, attributed to the pyelonephritis. Transthoracic echocardiography revealed a nonspecific cardiomyopathy phenotype with biatrial enlargement. Electrocardiography showed wide QRS complexes, bradycardia and synchronous dissociation between atrial and ventricular activity. While during the first examination the synchronisation appeared mostly fixed, with consistent PR intervals, on subsequent evaluations the PR intervals appeared to vary on a beat-tobeat basis. The diagnosis was of Isorhythmic Atrioventricular Dissociation (IAVD) type I, with an initial long phase of fixed synchronisation, mimicking an IAVD type II, and concomitant ventricular depolarisation with a left bundle branch block morphology
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇