Search results
Found 9803 matches for
Human-Like Epistemic Trust? A Conceptual and Normative Analysis of Conversational AI in Mental Healthcare.
The attribution of human concepts to conversational artificial intelligence (CAI) simulating human characteristics and conversation in psychotherapeutic settings presents significant conceptual and normative challenges. First, this article analyzes the concept of epistemic trust by identifying its problematic conditions when attributed to CAI, arguing for conceptual shift. We propose a conceptual, visual tool to navigate this shift. Second, three conceptualizations of AI are analyzed to understand their influence on the interpretation and evaluation of conceptual shift of epistemic trust and associated risks. We contrast two common AI conceptualizations from literature: a dichotomic account, distinguishing between AI's real and simulated abilities, and a relational account. Finally, we propose a novel approach: conceptualizing AI as a fictional character to combine their strengths, arguing for shifting focus from merely simulating human abilities to addressing CAI's actual strengths and weaknesses. The article sheds light on underlying theoretical assumptions that influence the ethical analysis of CAI.
Analysis of IDH1 and IDH2 mutations as causes of the hypermethylator phenotype in colorectal cancer.
The CpG island methylator phenotype (CIMP) occurs in many colorectal cancers (CRCs). CIMP is closely associated with global hypermethylation and tends to occur in proximal tumours with microsatellite instability (MSI), but its origins have been obscure. A few CRCs carry oncogenic (gain-of-function) mutations in isocitrate dehydrogenase IDH1. Whilst IDH1 is an established CRC driver gene, the low frequency of IDH1-mutant CRCs (about 0.5%) has meant that the effects and molecular covariates of those mutations have not been established. We first showed computationally that IDH2 is also a CRC driver. Using multiple public and in-house CRC datasets, we then identified IDH mutations at the hotspots (IDH1 codons 132 and IDH2 codons 140 and 172) frequently mutated in other tumour types. Somatic IDH mutations were associated with BRAF mutations and expression of mucinous/goblet cell markers, but not with KRAS mutations or MSI. All IDH-mutant CRCs were CIMP-positive, mostly at a high level. Cell and mouse models showed that IDH mutation was plausibly causal for DNA hypermethylation. Whilst the aetiology of hypermethylation generally remains unexplained, IDH-mutant tumours did not form a discrete methylation subcluster, suggesting that different underlying mechanisms can converge on similar final methylation phenotypes. Although further analysis is required, IDH mutations may be the first cause of hypermethylation to be identified in a common cancer type, providing evidence that CIMP and DNA methylation represent more than aging-related epiphenomena. Cautious exploration of mutant IDH inhibitors and DNA demethylating agents is suggested in managing IDH-mutant CRCs. © 2025 The Author(s). The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.
Absolute and relative risks of mental disorders in families: a Danish register-based study
Background: Relative risk estimates of familial aggregation of many types of mental disorders are available, but absolute risk estimates of familial aggregation of mental disorders remain sparse. The proportion of individuals who develop a mental disorder in the absence of the same disorder in a relative (non-familial cases) has not been examined. We aimed to create comprehensive risk estimates of the familial aggregation of mental disorders. Methods: In this prospective cohort study, we followed people of Danish origin between Jan 1, 1970, and Dec 31, 2021. We used Danish population-based registers to link individuals and their mental health across extended family pedigrees. These registers include the Danish Civil Registration System, the Danish Multi Generation Register, the Danish Psychiatric Central Research Register, and the Danish National Patient Register. Mental disorders investigated were substance use disorder, cannabis use disorder, alcohol use disorder, schizophrenia and related disorders, schizophrenia, schizoaffective disorder, mood disorders, bipolar disorder, single and recurrent depressive disorders (depression), personality disorder, borderline personality disorder, and antisocial personality disorder. We estimated lifetime risk (risk up to age 60 years), age-specific absolute risk, and relative risk for each mental disorder and type of affected relative (eg first, second, or third-degree relatives). We calculated heritability estimates and the proportion of non-familial cases. We involved people with related lived experience in the study design and implementation. Findings: A total of 3 048 583 individuals (1 486 132 [48·75%] females and 1 562 451 [51·25%] males) were followed up for 80 425 971 person-years. Individuals with a family member with a specific type of mental disorder had higher lifetime and relative risks of developing the same type of mental disorder. Both lifetime and relative risks were higher the closer the affected kinship. For example, the lifetime risk of depression was 15·48% (95% CI 15·31–15·65) in individuals with affected first-degree relatives, 13·50% (13·25–13·75) in individuals with affected second-degree relatives, 7·80% (7·76–7·84) in the general population, and 4·68% (4·65–4·71) in individuals without affected first-degree and second-degree relatives. The heritability for depression was 45·4% (95% CI 44·8–46·0) and the proportion of non-familial cases constituted 60·0% (95% CI 59·8–60·2). Interpretation: Individuals with family members with a mental disorder face increased risks of the same disorder. From a population perspective, most mental disorders occur in individuals without affected close relatives, thus highlighting the need for prevention strategies which target the entire population. Funding: Novo Nordisk Foundation. Translation: For the Danish translation of the abstract see Supplementary Materials section.
The Psychiatric Genomics Consortium: discoveries and directions.
Research by the Psychiatric Genomics Consortium (PGC) has advanced the discovery of common and rare genetic variations that contribute to the susceptibility to many psychiatric disorders and neurodevelopmental conditions. This Review reflects on major findings from the past 5 years of research by the PGC in five priority areas: discovery of common variants using genome-wide association studies; rare variation and its interplay with polygenic risk; using genetics to go beyond diagnostic boundaries; ascribing functional attributes to genomic discoveries; and developing and implementing processes for data sharing, outreach to various communities, and training. The insights gained in these domains frame the agenda for the next phase of PGC research. In addition to accelerating integrative findings of common and rare variants within, and across, multiple psychiatric disorders and neurodevelopmental conditions, the next phase will use multiple populations to elucidate genetic causes, integrate results with rapidly accumulating multimodal functional genomics data to gain mechanistic understanding, convert genetic findings to clinically actionable phenotypes, such as treatment response, and address the emerging use of polygenic scores. Together, these next steps will highlight the biological underpinnings of psychiatric disorders and neurodevelopmental conditions, which continue to contribute to global morbidity and mortality.
Augmentation of Cardiac Ischemic Geometry for Improving Machine Learning Performance in Arrhythmic Risk Stratification
AbstractVentricular arrhythmias frequently occur as a complication of myocardial infarction (MI), due to significant changes in the heart’s structure and electrophysiology. If left untreated, these alterations may lead to sudden cardiac death (SCD). It is therefore critical to evaluate risk prediction accurately in post-infarction patients to enable early intervention and improve patient outcomes. This work introduces a novel approach to improve arrhythmia risk assessment in post-infarction patients. We propose a new pipeline to build physiologically realistic image-based models of patient hearts, producing more realistic meshes compared to publicly available pipelines. We generate a library of 90 cardiac geometries of MI patients and use these cardiac models to estimate likelihood of reentry using electrophysiological (EP) simulations. However, due to the computationally expensive nature of this approach, we also introduce a data augmentation pipeline to train a machine learning (ML) model for risk stratification, enabling accurate and real-time prediction of the simulation outcomes. Our trained ML model achieved an accuracy of 88.0% and F1 score of 48%, with a prediction time of 0.01 seconds per case (compare with approximately 5 hours per case for EP simulations). In conclusion, the work presented here improved the accuracy of personalised biventricular geometries, introduced a novel data augmentation approach for scar distribution, and decreased prediction time of risk of arrhythmias post-MI by more than five orders of magnitude.
Molnupiravir or nirmatrelvir-ritonavir plus usual care versus usual care alone in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial.
BackgroundMolnupiravir and nirmatrelvir-ritonavir are oral antivirals that have shown efficacy in preventing disease progression in outpatients with COVID-19. We aimed to evaluate these treatments for patients hospitalised with COVID-19 pneumonia, for whom data on these antivirals are scarce.MethodsThe RECOVERY trial is a randomised, controlled, open-label, adaptive platform trial testing treatments for COVID-19. In this study we report the molnupiravir and nirmatrelvir-ritonavir comparisons from the RECOVERY trial. In each comparison, participants aged 18 years and older were randomly allocated (1:1) to the relevant antiviral (5 days of molnupiravir 800 mg twice daily or 300 mg nirmatrelvir and 100 mg ritonavir twice daily) in addition to usual care, or to usual care alone. The molnupiravir comparison was conducted at 75 hospitals in the UK, two in Nepal, and two in Indonesia; the nirmatrelvir-ritonavir comparison was conducted at 32 hospitals in the UK. Participants could take part in both comparisons. The primary outcome was 28-day mortality, and secondary outcomes were time to discharge alive from hospital and progression to invasive ventilation or death. Analysis was by intention to treat. Both comparisons were stopped because of low recruitment. This study is registered with ISRCTN, 50189673, and ClinicalTrials.gov, NCT04381936.FindingsFrom Jan 24, 2022, to May 24, 2023, 923 participants were recruited to the molnupiravir comparison (445 allocated to molnupiravir and 478 to usual care), and from March 31, 2022, to May 24, 2023, 137 participants were recruited to the nirmatrelvir-ritonavir comparison (68 allocated to nirmatrelvir-ritonavir and 69 to usual care). More than three-quarters of participants were vaccinated and had antispike antibodies at randomisation, and more than two-thirds were receiving other SARS-CoV-2 antivirals. In the molnupiravir comparison, 74 (17%) participants allocated to molnupiravir and 79 (17%) allocated to usual care died within 28 days (hazard ratio [HR] 0·93 [95% CI 0·68-1·28], p=0·66). In the nirmatrelvir-ritonavir comparison, 13 (19%) participants allocated to nirmatrelvir-ritonavir and 13 (19%) allocated to usual care died within 28 days (HR 1·02 [0·47-2·23], p=0·96). In neither comparison was there evidence of any difference in the duration of hospitalisation or the proportion of participants progressing to invasive ventilation or death.InterpretationAdding molnupiravir or nirmatrelvir-ritonavir to usual care was not associated with improvements in clinical outcomes. However, low recruitment meant a clinically meaningful benefit of treatment could not be ruled out, particularly for nirmatrelvir-ritonavir.FundingUK Research and Innovation (UK Medical Research Council), the National Institute for Health and Care Research, and the Wellcome Trust.
Machine learning in Alzheimer's disease genetics.
Traditional statistical approaches have advanced our understanding of the genetics of complex diseases, yet are limited to linear additive models. Here we applied machine learning (ML) to genome-wide data from 41,686 individuals in the largest European consortium on Alzheimer's disease (AD) to investigate the effectiveness of various ML algorithms in replicating known findings, discovering novel loci, and predicting individuals at risk. We utilised Gradient Boosting Machines (GBMs), biological pathway-informed Neural Networks (NNs), and Model-based Multifactor Dimensionality Reduction (MB-MDR) models. ML approaches successfully captured all genome-wide significant genetic variants identified in the training set and 22% of associations from larger meta-analyses. They highlight 6 novel loci which replicate in an external dataset, including variants which map to ARHGAP25, LY6H, COG7, SOD1 and ZNF597. They further identify novel association in AP4E1, refining the genetic landscape of the known SPPL2A locus. Our results demonstrate that machine learning methods can achieve predictive performance comparable to classical approaches in genetic epidemiology and have the potential to uncover novel loci that remain undetected by traditional GWAS. These insights provide a complementary avenue for advancing the understanding of AD genetics.
Longer scans boost prediction and cut costs in brain-wide association studies.
A pervasive dilemma in brain-wide association studies1 (BWAS) is whether to prioritize functional magnetic resonance imaging (fMRI) scan time or sample size. We derive a theoretical model showing that individual-level phenotypic prediction accuracy increases with sample size and total scan duration (sample size × scan time per participant). The model explains empirical prediction accuracies well across 76 phenotypes from nine resting-fMRI and task-fMRI datasets (R2 = 0.89), spanning diverse scanners, acquisitions, racial groups, disorders and ages. For scans of ≤20 min, accuracy increases linearly with the logarithm of the total scan duration, suggesting that sample size and scan time are initially interchangeable. However, sample size is ultimately more important. Nevertheless, when accounting for the overhead costs of each participant (such as recruitment), longer scans can be substantially cheaper than larger sample size for improving prediction performance. To achieve high prediction performance, 10 min scans are cost inefficient. In most scenarios, the optimal scan time is at least 20 min. On average, 30 min scans are the most cost-effective, yielding 22% savings over 10 min scans. Overshooting the optimal scan time is cheaper than undershooting it, so we recommend a scan time of at least 30 min. Compared with resting-state whole-brain BWAS, the most cost-effective scan time is shorter for task-fMRI and longer for subcortical-to-whole-brain BWAS. In contrast to standard power calculations, our results suggest that jointly optimizing sample size and scan time can boost prediction accuracy while cutting costs. Our empirical reference is available online for future study design ( https://thomasyeolab.github.io/OptimalScanTimeCalculator/index.html ).
An atlas of trait associations with resting-state and task-evoked human brain functional organizations in the UK Biobank.
Functional magnetic resonance imaging (fMRI) has been widely used to identify brain regions linked to critical functions, such as language and vision, and to detect tumors, strokes, brain injuries, and diseases. It is now known that large sample sizes are necessary for fMRI studies to detect small effect sizes and produce reproducible results. Here we report a systematic association analysis of 647 traits with imaging features extracted from resting-state and task-evoked fMRI data of more than 40,000 UK Biobank participants. We used a parcellation-based approach to generate 64,620 functional connectivity measures to reveal fine-grained details about cerebral cortex functional organizations. The difference between functional organizations at rest and during task was examined, and we have prioritized important brain regions and networks associated with a variety of human traits and clinical outcomes. For example, depression was most strongly associated with decreased connectivity in the somatomotor network. We have made our results publicly available and developed a browser framework to facilitate the exploration of brain function-trait association results (http://fmriatlas.org/).
Evaluating functional brain organization in individuals and identifying contributions to network overlap.
Individual differences in the spatial organization of resting-state networks have received increased attention in recent years. Measures of individual-specific spatial organization of brain networks and overlapping network organization have been linked to important behavioral and clinical traits and are therefore potential biomarker targets for personalized psychiatry approaches. To better understand individual-specific spatial brain organization, this paper addressed three key goals. First, we determined whether it is possible to reliably estimate weighted (non-binarized) resting-state network maps using data from only a single individual, while also maintaining maximum spatial correspondence across individuals. Second, we determined the degree of spatial overlap between distinct networks, using test-retest and twin data. Third, we systematically tested multiple hypotheses (spatial mixing, temporal switching, and coupling) as candidate explanations for why networks overlap spatially. To estimate weighted network organization, we adopt the Probabilistic Functional Modes (PROFUMO) algorithm, which implements a Bayesian framework with hemodynamic and connectivity priors to supplement optimization for spatial sparsity/independence. Our findings showed that replicable individual-specific estimates of weighted resting-state networks can be derived using high-quality fMRI data within individual subjects. Network organization estimates using only data from each individual subject closely resembled group-informed network estimates (which was not explicitly modeled in our individual-specific analyses), suggesting that cross-subject correspondence was largely maintained. Furthermore, our results confirmed the presence of spatial overlap in network organization, which was replicable across sessions within individuals and in monozygotic twin pairs. Intriguingly, our findings provide evidence that overlap between 2-network pairs is indicative of coupling. These results suggest that regions of network overlap concurrently process information from both contributing networks, potentially pointing to the role of overlapping network organization in the integration of information across multiple brain systems.
“Looking at the Big Picture”: A Qualitative Study of Ethics in Science Communication and Engagement
Ethical issues arise in many communication and engagement settings. Such issues can, however, fall into the gaps between what is seen as “research” and what is seen as “dissemination.” Semi-structured interviews (n = 17) and focus groups (n = 2) with researchers and science communication and public engagement specialists at U.K. academic institutions and in practice settings suggest that while normative principles for ethical science communication remain fluid, ethical questions are often an area of considerable reflection for those communicating, particularly when they reflect wider social issues and different people in the process: communities, researchers, and institutions.