Valentina Giunchiglia

The mechanism by which idiopathic normal pressure hydrocephalus (iNPH) causes cognitive impairment, gait dysfunction and urinary incontinence is unclear. The key radiological feature is ventriculomegaly, and symptoms can be treated with surgery. We hypothesise that ventricular expansion causes clinical effects by stretching and/or compressing white matter tracts. We define disease-specific white matter abnormalities, explore their relationship to clinical features and test whether they can be explained by a computational model of ventricular expansion in iNPH. In this observational study, diffusion tensor imaging with Tract-Based Spatial Statistics analysis was used to investigate white matter abnormalities in 22 patients with iNPH (14 males, mean age=71.91, SD=7.06), 28 with Alzheimer’s disease (AD) (18 males, mean age=74.80), 23 with moderate-severe traumatic brain injury (TBI) (15 males, mean age=69.87) and 30 healthy controls (16 males, mean age=73.80). We developed a finite-element biomechanical model of hydrocephalus, which predicts patterns of white matter stretch and compression generated to deform from healthy to iNPH brain, and studied their effect on the extracted diffusion properties. Increased ventricular size relative to controls was found in both iNPH, AD and TBI. However, iNPH showed a unique pattern of increased and decreased fractional anisotropy (FA) in distinct sections of the corticospinal tract and corona radiata compared to AD, TBI and controls. Biomechanical modelling revealed that stretch in these tracts was associated with increased FA, driven by increased axial diffusivity (AxD) with little change or a reduction in radial diffusivity (RD). In contrast, compression was associated with reduced FA, resulting from increases in both AxD and RD. Parcel-based correlation analysis showed that strain predicted by the biomechanical model in these tracts was strongly and positively correlated with FA and negatively correlated with RD. The corpus callosum experienced greater stretch along its entire length than the corona radiata and corticospinal tract, yet it demonstrated the largest reduction in FA, driven by increases in both AxD and RD. In iNPH, processing speed impairments were associated with increased AxD in the corpus callosum and AxD and RD in the corona radiata. Higher urinary incontinence correlated with reduced FA in the corona radiata and corticospinal tract. iNPH is characterised by a distinct biomechanical substrate, in which predictable patterns of stretch and compression associated with ventricular expansion lead to clinically relevant white matter damage. These diffusion abnormalities distinguish iNPH from AD and TBI and may serve as potential biomarkers. Longitudinal studies of patients undergoing CSF drainage are warranted.

AI agents promise to empower biomedical discovery, but realizing this promise requires the ability to complete transparent, long-horizon analyses using tools. Agents must make intermediate decisions explicit, and validate each decision and output against data and tool constraints as the analysis unfolds. We present M edea , an AI agent that takes an omics objective and executes a transparent multi-step analysis using tools. M edea comprises four modules: research planning with context and integrity verification, code execution with pre- and post-run checks, literature reasoning with evidence-strength assessment, and a consensus stage that reconciles evidence across datasets, tools, and literature. M edea uses 20 tools spanning single-cell and bulk transcriptomic datasets, cancer vulnerability maps, pathway knowledge bases, and machine learning models. We evaluate M edea across 5,679 analyses in three open-ended domains: target identification across five diseases and cell type contexts (2,400 analyses), synthetic lethality reasoning in seven cell lines (2,385 analyses), and immunotherapy response prediction in bladder cancer (894 patient analyses). In evaluations that vary large language models, tool sets, omics objectives, and agentic modules, M edea improves the performance of existing approaches by up to 46% for target identification, 22% for synthetic lethality, and 24% for immunotherapy response prediction, while maintaining low failure rates and calibrated abstention. M edea shows that verification-aware AI agents improve performance by producing transparent analyses, not simply more efficient workflows.

Large public datasets of the human microbiome now exist but combining them for large-scale analysis is difficult due to a lack of standardization. We developed curatedMetagenomicData (cMD) 3, a uniformly processed collection of over 22,000 human microbiome samples with manually curated metadata from 94 studies and 42 countries. This large and diverse resource allows for meta-analysis of the links between microbes and human health. Through meta-analysis, we identified hundreds of microbial species and thousands of microbial functions significantly associated with a person’s sex, age, body mass index, and disease status, and catalog these as references. We developed an “oral enrichment score” (OES) based on the relative abundance of bacteria typically found in the oral cavity and not in the gut. Higher OES in the gut is a consistent feature in individuals with disease, suggesting that the relative abundance of oral bacteria in the gut is a simple and quantifiable signal of altered microbiome health. These analyses identify modest but widely shared patterns in human microbiomes, serving as a reproducible and readily updatable reference.

Long COVID remains poorly understood and previous studies have mainly focused on describing symptoms or single-system health deficits in severe cases. In the CONVALESCENCE case-control study, 306 community-based participants (aged 57 [IQR 41–63], 80% female) were enrolled 2.0 [1.7, 2.5] years post-infection and were monitored for 12 months using a smartwatch and apps to assess mental health, cognitive function, physical activity, exercise capacity, cardiorespiratory function, autonomic function and sleep. Compared to controls (N = 152), long COVID cases with fatigue-cluster symptoms (N = 50) presented small persistent deficits across all systems with no evidence of recovery over 12 months. Participants with fatigue-cluster symptoms not attributable to long COVID (N = 28) showed a similar trend, while participants with prior long COVID but no fatigue-cluster symptoms (N = 76) were not different from controls. The lack of significant recovery between years two and three post-infection demonstrates protracted multi-system deficits for fatigue-cluster cases and highlights ongoing needs to better understand and manage long COVID.

Background: Cognitive assessments are essential for detecting and monitoring cognitive changes in neurological populations. Compared to standard pen‐and‐paper tests, online cognitive tasks offer a more accessible, scalable, repeatable and cost‐effective approach to assessment. Cognitron is an online cognitive assessment platform with previously demonstrated validity and reliability (1). This study aims to derive meaningful insights into the application of online and computerized assessments in older age cohorts by examine the usability of Cognitron tasks with the MRC National Survey of Health and Development (NSHD; the 1946 British birth cohort).Method: 1753 members of the NSHD cohort (all aged 77) were invited to complete a battery comprising 13 Cognitron tasks, spanning multiple cognitive domains, on their home devices. Each task generated measures of accuracy and reaction time. We examined data quality, completion rates, and levels of engagement and compliance. Qualitative data from email (N = 200) and telephone (N = 78) queries were analysed to determine insights into participant experiences.Result: 990 members (56.47%) consented to participate in the study, of whom 813 attempted the battery (46.34%). 739/813 (90.89%) completed all cognitive tests (average completion time = 40 minutes). 8 tasks (61.53%) were completed by all participants. 4 tasks had 90% of trials (Table 1). The qualitative analysis of the reasons participants contacted the study by email or telephone identified five key themes: (1) technical issues (N = 137), (2) general queries (N = 88), (3) reasons for withdrawing (N = 72), (4) providing feedback (N = 43) and (5) subjective reports (N = 28) (Figure 2).Conclusion: Cognitron assessment was characterised by high completion, engagement and compliance rates in the NSHD cohort, suggesting that online cognitive assessments have potential to monitor cognitive changes in older age cohorts, including those at risk of developing dementia.Reference(1) Del Giovane, Martina, et al. (2023) EClinicalMedicine 59 https://doi.org/10.1016/j.eclinm.2023.101980

Online cognitive tasks are gaining traction as scalable and cost-effective alternatives to traditional supervised assessments. However, variability in peoples’ home devices, visual and motor abilities, and speed-accuracy biases confound the specificity with which online tasks can measure cognitive abilities. To address these limitations, we developed IDoCT (Iterative Decomposition of Cognitive Tasks), a method for estimating domain-specific cognitive abilities and trial-difficulty scales from task performance timecourses in a data-driven manner while accounting for device and visuomotor latencies, unspecific cognitive processes and speed-accuracy trade-offs. IDoCT can operate with any computerised task where cognitive difficulty varies across trials. Using data from 388,757 adults, we show that IDoCT successfully dissociates cognitive abilities from these confounding factors. The resultant cognitive scores exhibit stronger dissociation of psychometric factors, improved cross-participants distributions, and meaningful demographic’s associations. We propose that IDoCT can enhance the precision of online cognitive assessments, especially in large scale clinical and research applications.

INTRODUCTION: Online assessments are scalable and cost-effective for detecting cognitive changes, especially in elderly cohorts with limited mobility and higher vulnerability to neurological conditions. However, determining the uptake, adherence, and usability of these assessments in older adults, who may have less experience with mobile devices is crucial. METHODS: 1,776 members (aged 77) of the MRC National Survey of Health and Development (NSHD) were invited to complete 13 online cognitive tasks. Adherence was measured through task compliance, while uptake (consent, attempt, completion) was linked to health and sociodemographic factors. Usability was evaluated through qualitative feedback. RESULTS: This study’s consent (56.9%), attempt (80.5%), and completion (88.8%) rates are comparable to supervised NSHD sub-studies. Significant predictors of uptake included education, sex, handedness, cognitive scores, weight, smoking, alcohol consumption, and disease burden. DISCUSSION: With key recommendations followed, online cognitive assessments are feasible, with good adherence, and usability in older adults.

Automated online cognitive assessments are set to revolutionise clinical research and healthcare. However, their applicability for Parkinson’s Disease (PD) and REM Sleep Behavioural Disorder (RBD), a strong PD precursor, is underexplored. Here, we developed an online battery to measure early cognitive changes in PD and RBD. Evaluating 19 candidate tasks showed significant global accuracy deficits in PD (0.65 SD, p = 0.003) and RBD (0.45 SD, p = 0.027), driven by memory, language, attention and executive underperformance, and global reaction time deficits in PD (0.61 SD, p = 0.001). We identified a brief 20-min battery that had sensitivity to deficits across these cognitive domains while being robust to the device used. This battery was more sensitive to early-stage and prodromal deficits than the supervised neuropsychological scales. It also diverged from those scales, capturing additional cognitive factors sensitive to PD and RBD. This technology offers an economical and scalable method for assessing these populations that can complement standard supervised practices.

Automated online and App-based cognitive assessment tasks are becoming increasingly popular in large-scale cohorts and biobanks due to advantages in affordability, scalability, and repeatability. However, the summary scores that such tasks generate typically conflate the cognitive processes that are the intended focus of assessment with basic visuo-motor speeds, testing device latencies, and speed-accuracy tradeoffs. This lack of precision presents a fundamental limitation when studying brain-behaviour associations. Previously, we developed a novel modelling approach that leverages continuous performance recordings from large-cohort studies to achieve an iterative decomposition of cognitive tasks (IDoCT), which outputs data-driven estimates of cognitive abilities, and device and visuo-motor latencies, whilst recalibrating trial-difficulty scales. Here, we further validate the IDoCT approach with UK BioBank imaging data. First, we examine whether IDoCT can improve ability distributions and trial-difficulty scales from an adaptive picture-vocabulary task (PVT). Then, we confirm that the resultant visuo-motor and cognitive estimates associate more robustly with age and education than the original PVT scores. Finally, we conduct a multimodal brain-wide association study with free-text analysis to test whether the brain regions that predict the IDoCT estimates have the expected differential relationships with visuo-motor versus language and memory labels within the broader imaging literature. Our results support the view that the rich performance timecourses recorded during computerised cognitive assessments can be leveraged with modelling frameworks like IDoCT to provide estimates of human cognitive abilities that have superior distributions, re-test reliabilities, and brain-wide associations.