The global rise in dementia necessitates scalable cognitive assessments that can evolve to serve both clinical and research applications. We present the Oxford Cognitive Testing Portal (OCTAL), a remote, browser-based platform providing performance metrics for memory, attention, visuospatial and executive function domains. Four validation studies (N = 1664) confirmed cross-cultural applicability, lifespan sensitivity and clinical utility. Task performance was equivalent in English- and Chinese-speaking younger adults and mapped domain-specific ageing trajectories in mid- to late-adulthood. In a memory-clinic cohort (N = 194), 5-minute OCTAL screen distinguished patients with Alzheimer’s disease dementia from subjective cognitive decline (AUC = 0.92), matching a standard paper-based test, while a 20-minute subset surpassed this (AUC = 0.97; p = 0.04). Test-retest reliability was very good (ICC ≥ 0.79; N = 118). OCTAL enables remote assessment for large-scale research and screening, with an open, modular architecture that makes it a uniquely sustainable and evolvable tool for the research community.

Apathy is a highly prevalent and disabling neuropsychiatric syndrome, but its multi-dimensional structure is a challenge for progress towards better identification and treatment. A crucial unresolved question is whether social disengagement reflects a distinct deficit in social motivation or a by-product of diminished initiative or emotional blunting. Previous studies have been constrained by modest sample sizes and limited use of apathy-specific instruments or phenotypically narrow cohorts. Here, we analysed item-level data from 11,243 individuals recruited across multiple centres, including 1154 neurological patients with Alzheimer’s disease, Parkinson’s disease, frontotemporal dementia, autoimmune encephalitis and small vessel disease, alongside people with depression and healthy adults. Across exploratory and confirmatory factor analyses, symptom-level network modelling, and lifespan analyses, social apathy consistently emerged as a coherent and separable dimension. This pattern was preserved across health, psychiatric, and neurocognitive cohorts, from adolescence through late life. Recognising social apathy as an independent domain reframes a central aspect of mental health—the motivation to connect, care, and act for others—and provides a foundation for more precise assessment and for interventions targeting both social and neurobiological mechanisms.

The brain continuously integrates rapidly changing visual input across eye movements to maintain stable perception, yet the precise mechanisms underpinning dynamic working memory and how these break down in brain diseases remain unclear. We developed a novel eye-tracking paradigm and computational models to investigate how spatial and colour information are updated across saccades in the human brain. Our findings reveal that saccades selectively impair spatial but not colour memory. Computational modelling identified that spatial representations are maintained in a dual eye-centred frame of reference which is actively updated by a noisy memory of saccades but is vulnerable to interference. Using this model, we found that specific mechanistic failures in initial encoding and memory decay, rather than the saccadic updating process itself, account for spatial working memory deficits in Alzheimer’s and Parkinson’s disease. These results provide a mechanistic understanding of how dynamic spatial memory operates in health and its disruption in neurodegenerative disorders.

White matter microstructural abnormalities are increasingly found to be associated with cognitive impairment in Alzheimer’s Disease (AD). Here, we investigated the relationship between visual short-term memory (VSTM) performance, measured using a digital cognitive task, and integrity of brain white matter tracts. 52 AD and 60 age-matched healthy controls were recruited from the Oxford Cognitive Disorders Clinic. An established digital VSTM test – the Oxford Memory Task (OMT) – was used to measure several key memory metrics including: Identification Accuracy (percentage of correctly identified items), Target detection (probability of correctly identifying the target) and Misbinding (erroneously localizing an item to the remembered location of another item in memory). TBSS (Tract-Based Spatial Statistics) was then applied to investigate in which regions of the brain microstructural disruption of physiological diffusivity correlated with behavioural performance. A key common area, comprising the left optic radiation, forceps major and middle longitudinal fasciculus (MLF), was associated with performance across several VSTM metrics in patients with AD. In addition, misbinding was linked to the left MLF (part III), left inferior fronto-occipital fasciculus and left vertical occipital fasciculus. Microstructural disruption associated with target detection was additionally associated with superior longitudinal fasciculus, and Identification Accuracy with left superior thalamic radiation in AD. These findings reveal a common shared area of altered diffusivity in AD patients linked to global impairment in VSTM, while distinct types of memory errors are associated with disruption of additional contributions of selective white matter tracts. Key points ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement This research was supported by funding from the Wellcome Trust (206330/Z/17/Z) and National Institute for Health and Care Research (NIHR) Oxford Health Biomedical Research Centre. Y.A.T. was supported by a PhD scholarship by the Friedrich-Ebert-Stiftung. I.M.I was supported by the University of Oxford and the University of Malaya. The funder played no role in study design, data collection, analysis and interpretation of data, or the writing of this manuscript. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: Ethical approval was granted by the University of Oxford ethics committee (IRAS ID: 248379, Ethics Approval Reference: 18/SC/0448). I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes De-identified data supporting this study may be shared based on reasonable written requests to the corresponding author. Access to de-identified data will require a Data Access Agreement and IRB clearance, which will be considered by the institutions who provided the data for this research.

The hippocampus is increasingly recognized for its role in working memory representations and goal-directed behavior, yet whether these functions share common neural substrates remains unclear. We investigated short-term memory performance in 27 patients (21 males and 6 females) with LGI1-antibody limbic encephalitis, a condition with predominant hippocampal involvement, compared to 27 age- and gender-matched healthy controls, using an object-location continuous report task with Bayesian mixture-modeling. Patients exhibited predominantly elevated misbinding errors-incorrectly associating objects with non-target locations-with unaltered guessing and only marginally reduced precision. Critically, apathy was positively associated with misbinding exclusively in patients, independent of depression and global cognition. Neuroimaging in a subset of patients (n = 12) revealed reduced hippocampal volumes and preliminary evidence for diminished hippocampal-medial prefrontal connectivity, which was associated with both higher apathy and increased misbinding at short retention intervals. These findings demonstrate that hippocampal dysfunction produces convergent deficits in memory binding and motivation, potentially reflecting disruption of a common brain mechanism linking working memory processing to goal-directed behavior.

IntroductionDementias, including those caused by Alzheimer's Disease (AD), are a leading global cause of death, necessitating improvements in early detection and development of more effective disease-modifying therapies. Well-validated pen-and-paper measures serve as the primary method used for cognitive assessment in research and clinical trials in AD. However, these suffer from rater error, infrequent “snapshot” bias, and have limited sensitivity to early-stage pathology, which presents challenges for measuring efficacy in prevention trials. Recent technological advances offer the potential to measure subtle cognitive changes and account for day-to-day variability through real-world data collection. The Cumulus Neuroscience NeuLogiq® platform (“the platform”), is comprised of a wireless electroencephalography (EEG) headset, tablet-based cognitive tasks based on well-established paradigms which were designed to be user-friendly based on patient panel inputs, third party integrations (mood, speech and sleep assessments) and cloud-based analytics. This platform has demonstrated utility in healthy populations and may enable objective, frequent, and patient-centered disease tracking in AD and other CNS disorders.MethodsThis paper presents findings from a 52-week study involving individuals with mild AD dementia and healthy controls. Analyses focus on usability (e.g., ease of use ratings and reported technical issues) and feasibility (e.g., adherence; withdrawal rates) of using the platform, unsupervised, in the real-world at home setting and exploring how baseline cognitive status and demographic factors impact platform usability.ResultsLongitudinal study data after 12 months offer valuable insights into the feasibility of the platform for patients with mild AD enrolled in long-term studies, such as clinical trials. The participants' high adherence to the protocol underscores the practicality of utilizing the platform in this context. Although participants with dementia reported lower confidence levels (31.6%, N = 18) and encountered some minor technical challenges during the initial home setup (accounting for 16.1% of issues reported in Stage 1), they nonetheless demonstrated strong engagement, achieving an overall adherence rate of 77% across the 52-week study protocol.DiscussionThis demonstrates that even participants with dementia remain able and willing to use the EEG headset and complete tablet-based cognitive tasks over a year, from the comfort of their homes.

Background Digital cognitive testing and plasma biomarkers have emerged as key players for population screening and measuring drug efficacy in clinical trials for Alzheimer’s Disease (AD). Understanding test-retest reliability of these tools is essential to estimate meaningful effects of an intervention over time. Whilst some data exist on variability of plasma and digital biomarkers at very short time frames or annually, studies conducted at intermediate time scales are lacking. Data on impact on this variability on clinical decision-making are also rare. Methods Paired testing of plasma biomarkers and digital cognition was performed at 0, 3 and 6 months in a cohort of 92 individuals (55 cognitively unimpaired healthy controls (HC) and 37 patients with early AD dementia (AD)). Plasma biomarkers examined included pTau217, pTau181, NfL, GFAP and the Aβ42/40 ratio. OCTAL (Oxford Cognitive Testing Portal) digital cognitive testing platform was used to measure cognition. Results pTau181 emerged as the plasma biomarker with worst test-retest reliability and was the only measure where repeating the sample improved accuracy in group discrimination. None of the digital measures showed further improvement in diagnostic accuracy if repeated over time, reflecting high reliability. Individual metrics showed different reliability profiles based on the index used to measure variability (Coefficient of variation or Intraclass Correlation Coefficient) and varied within groups (HC and AD). Conclusion Test-retest reliability measures are essential for interpretation of meaningful effects over time for plasma and digital outcomes.

Plasma biomarkers can detect the presence of Alzheimer's disease (AD) when cognitive symptoms have not yet emerged. However, measuring cognitive function remains essential for large scale population screening, monitoring of disease progression and response to treatment. Despite the availability of several digital platforms, data on sensitivity and specificity of online testing in discriminating between different forms of dementia is currently lacking. 391 participants (31 subjective cognitive decline, 28 mild cognitive impairment, 88 Alzheimer's disease dementia, 18 Lewy body disease, 9 Corticobasal syndrome, 17 Frontotemporal dementia and 201 age-matched controls) were recruited from the Oxford Centre for Cognitive Disorders and other memory centres across the UK taking part in the FAST study. They were tested on a fully remote online cognitive assessment tool, the Oxford Cognitive Testing Portal (OCTAL: https://octalportal.com ), including a visual short-term memory (Oxford Memory Test, OMT) and Trail Making Task (TMT); see Figure 1 for task schematics and metrics’ description. Plasma pTau217 was measured using the Alamar platform. OCTAL's metrics captured different stages of the disease as well as discriminated between different forms of dementia (Figure 2). Identification accuracy on OMT was lower in patients with MCI as well as other primary dementias compared to controls. Absolute Localization Error could also distinguish between MCI and AD dementia patients. Localization time was significantly higher only in patients with AD and not other forms of dementia. Trail making test performance was lower in Lewy body disease patients compared to all other groups. pTau217 had the highest accuracy as single biomarker in discriminating between AD dementia and healthy controls, but adding digital biomarkers significantly improved diagnostic accuracy (Z = -2.0884, p -value = 0.037), (Figure 3). These findings highlight the potential of the OCTAL platform for sensitive, and specific stratification of patients’ performance in a typical memory clinic population. They also emphasize the utility of combining plasma biomarkers and digital cognitive tests to improve diagnostic accuracy in AD. This fully remote platform provides a scalable approach for screening, clinical stratification and monitoring of patients with AD and other dementias.

Understanding the cognitive trajectory of a neurological disease can provide important insight on underlying mechanisms and disease progression. Cognitive impairment is now well established as beginning many years before the diagnosis of Alzheimer's disease, but pre-diagnostic profiles are unclear for other neurological conditions that may be associated with cognitive impairment. We analysed data from the prospective UK Biobank cohort with study baseline assessment performed between 2006 and 2010 and participants followed until 2021. We examined data from 497 252 participants, aged between 38 and 72 years at baseline, with an imaging sub-sample of 42 468 participants. Using time-to-diagnosis and time-from-diagnosis data in relation to time of assessment, we compared a continuous measure of executive function and magnetic resonance imaging brain measures of total grey matter (GM) and hippocampal volume in individuals with ischaemic stroke, focal epilepsy, Parkinson's disease, multiple sclerosis, motor neurone disease (amyotrophic lateral sclerosis) and migraine. Of the 497 252 participants [226 206 (45.5%) men, mean (SD) age, 57.5(8.1) years], 12 755 had ischaemic stroke, 6758 had a diagnosis of focal epilepsy, 3315 had Parkinson's disease, 2315 had multiple sclerosis, 559 had motor neurone disease and 18 254 had migraine either at study baseline or diagnosed during the follow-up period. Apart from motor neurone disease, all conditions had lower pre-diagnosis executive function compared to controls (assessment performed median 7.4 years before diagnosis). At a group level, focal epilepsy and multiple sclerosis showed a gradual worsening in executive function up to 15 years prior to diagnosis, while ischaemic stroke was characterised by a modest decline for a few years followed by a substantial reduction at the time of diagnosis. By contrast, participants with migraine showed a mild reduction in pre-diagnosis cognition compared to controls which improved following clinical diagnosis. Pre-diagnosis MRI GM volume was lower than controls for stroke, Parkinson's disease and multiple sclerosis (scans performed median 1.7 years before diagnosis), while other conditions had lower volumes post-diagnosis. These cognitive trajectory models reveal disease-specific temporal patterns at a group level, including a long cognitive prodrome associated with focal epilepsy and multiple sclerosis. The findings may help to prioritise risk management of individual diseases and inform clinical decision-making.

BACKGROUND:Plasma biomarkers can detect the presence of Alzheimer's disease (AD) when cognitive symptoms have not yet emerged. However, measuring cognitive function remains essential for large scale population screening, monitoring of disease progression and response to treatment. Despite the availability of several digital platforms, data on sensitivity and specificity of online testing in discriminating between different forms of dementia is currently lacking. METHOD:391 participants (31 subjective cognitive decline, 28 mild cognitive impairment, 88 Alzheimer's disease dementia, 18 Lewy body disease, 9 Corticobasal syndrome, 17 Frontotemporal dementia and 201 age-matched controls) were recruited from the Oxford Centre for Cognitive Disorders and other memory centres across the UK taking part in the FAST study. They were tested on a fully remote online cognitive assessment tool, the Oxford Cognitive Testing Portal (OCTAL: https://octalportal.com), including a visual short-term memory (Oxford Memory Test, OMT) and Trail Making Task (TMT); see Figure 1 for task schematics and metrics' description. Plasma pTau217 was measured using the Alamar platform. RESULT:OCTAL's metrics captured different stages of the disease as well as discriminated between different forms of dementia (Figure 2). Identification accuracy on OMT was lower in patients with MCI as well as other primary dementias compared to controls. Absolute Localization Error could also distinguish between MCI and AD dementia patients. Localization time was significantly higher only in patients with AD and not other forms of dementia. Trail making test performance was lower in Lewy body disease patients compared to all other groups. pTau217 had the highest accuracy as single biomarker in discriminating between AD dementia and healthy controls, but adding digital biomarkers significantly improved diagnostic accuracy (Z = -2.0884, p-value = 0.037), (Figure 3). CONCLUSION:These findings highlight the potential of the OCTAL platform for sensitive, and specific stratification of patients' performance in a typical memory clinic population. They also emphasize the utility of combining plasma biomarkers and digital cognitive tests to improve diagnostic accuracy in AD. This fully remote platform provides a scalable approach for screening, clinical stratification and monitoring of patients with AD and other dementias.

The brain continuously integrates rapidly changing visual input across eye movements to maintain stable perception, yet the precise mechanisms underpinning dynamic working memory and how these break down in brain diseases remain unclear. We developed a novel eye-tracking paradigm and computational models to investigate how spatial and colour information are updated across saccades. Our findings reveal that saccades selectively impair spatial but not colour memory. Computational modelling identified that spatial representations are maintained in a dual eye-centred frame of reference which is actively updated by a noisy memory of saccades but is vulnerable to interference. Using this model, we found that specific mechanistic failures in initial encoding and memory decay, rather than the saccadic updating process itself, account for spatial working memory deficits in Alzheimer’s and Parkinson’s disease. These results provide a mechanistic understanding of how dynamic spatial memory operates in health and its disruption in neurodegenerative disorders.

Plasma biomarkers have emerged as a promising tool to detect the presence of Alzheimer’s disease (AD) when cognitive symptoms have not yet emerged. However, there is also a pressing need to detect and track subtle cognitive change at the preclinical stage of AD for population screening purposes and to monitor disease progression at scale. A potential solution is remote cognitive assessment, yet it is still not extensively employed. 114 participants (37 AD, 22 subjective cognitive impairment (SCI) patients and 55 age-matched controls) were recruited from the Oxford Centre for Cognitive Disorders. They were tested on a newly developed, fully remote online cognitive assessment tool, Oxford Cognitive Testing Portal (OCTAL), which hosts a wide range of validated cognitive tasks, such as the Rey-Osterrieth Complex Figure (ROCF), Trail Making Task (TMT), Digital Symbol Substitution task (DSST)), Corsi Block Task (CORSI), as well as novel visual short-term memory (Oxford Memory task) and visual long-term memory (Object-in-Scene task) tasks (Figure 1). All participants underwent standard in-person cognitive testing, i.e. Addenbrooke's Cognitive Examination-III (ACE). Plasma p-tau181, GFAP, NFL, Aβ42/40 ratio were also measured. Logistic regression was used for group classification. Performance on OCTAL was able to discriminate between SCI from healthy controls with an AUC of 0.78 (Figure 2a), statistically outperforming standard neuropsychological testing (ACE), which had an AUC of 0.65. Combining plasma biomarkers to OCTAL further increased diagnostic accuracy, reaching an AUC of 0.82 in group classification. OCTAL also outperformed ACE in distinguishing individuals with AD from SCI (Figure 2b), where adding plasma biomarkers to performance at OCTAL achieved a perfect separation (AUC of 1) between the groups. These findings demonstrate the potential of OCTAL for widespread, cost-effective cognitive testing. They also emphasise the utility of combining plasma biomarkers and digital cognitive tests to improve diagnostic accuracy across different stages of the disease. These accessible tools could pave the way to more scalable protocols for screening, stratification and monitoring of patients with preclinical AD.

Evaluate the association between functional recovery and a panel of specific neuronal biomarkers, in a cohort of stroke patients. Serum levels of neuronal specific enolase (NSE), neurofilament light chain (NfL), brain derived neurotrophic factor (BDNF), amyloid-β42 and β40 peptides (Aβ42/Aβ40 ratio) and total tau (t-tau), were measured in 20 patients within one month after stroke event (baseline, T0). After six weeks of extensive multimodal cognitive and motor rehabilitation (T1), levels of each biomarker were correlated with changes in clinical scales for disability and mobility (Barthel Index (BI), Rivermead Mobility Index (RMI), Functional Ambulation Categories (FAC), Fugl Mayer Assessment Upper Extremity (FMA). Linear regression was performed to predict changes in clinical scales during follow up, according to baseline biomarkers levels. NSE at T0 was a significant predictor of improvement in FAC and RMI, where the higher the NSE concentration, the smaller the improvement. Therefore, baseline NSE explained 39

Background Functional cognitive disorder (FCD) poses a diagnostic challenge due to its resemblance to other neurocognitive disorders and limited biomarker accuracy. We aimed to develop a new diagnostic checklist to identify FCD versus other neurocognitive disorders.Methods The clinical checklist was developed through mixed methods: (1) a literature review, (2) a three-round Delphi study with 45 clinicians from 12 countries and (3) a pilot discriminative accuracy study in consecutive patients attending seven memory services across the UK. Items gathering consensus were incorporated into a pilot checklist. Item redundancy was evaluated with phi coefficients. A briefer checklist was produced by removing items with >10% missing data. Internal validity was tested using Cronbach’s alpha. Optimal cut-off scores were determined using receiver operating characteristic curve analysis.Results A full 11-item checklist and a 7-item briefer checklist were produced. Overall, 239 patients (143 FCD, 96 non-FCD diagnoses) were included. The checklist scores were significantly different across subgroups (FCD and other neurocognitive disorders) (F(2, 236)=313.3, p<0.001). The area under the curve was excellent for both the full checklist (0.97, 95% CI 0.95 to 0.99) and its brief version (0.96, 95% CI 0.93 to 0.98). Optimal cut-off scores corresponded to a specificity of 97% and positive predictive value of 91% for identifying FCD. Both versions showed good internal validity (>0.80).Conclusions This pilot study shows that a brief clinical checklist may serve as a quick complementary tool to differentiate patients with neurodegeneration from those with FCD. Prospective blind large-scale validation in diverse populations is warranted.Cite Now

Gene regulation in humans extends beyond the four letter genetic code. Cytosine methylation, in particular, functions as a critical epigenetic switchboard, dynamically programming cellular identity, adapting gene expression in response to environmental cues, and underpinning the onset and progression of numerous diseases. Here we present Pleiades, a series of whole-genome epigenetic foundation models spanning three sizes: 90M, 600M, and 7B parameters. Pleiades is trained upon an extensive proprietary dataset of methylated and unmethylated human DNA sequences totalling 1.9T tokens. We introduce alignment embeddings and stacked hierarchical attention techniques to provide precise epigenetic modelling without the need for extended context lengths. Collectively, these advances enable Pleiades to perform a diverse range of downstream biological and clinical tasks, including nucleotide-level regulatory prediction, realistic generation of cell-free DNA fragments and fragment-level celltype-of-origin classification, within a unified and scalable computational framework. We specifically apply Pleiades to the early detection of real-world cohorts of clinical Alzheimer’s disease and Parkinson’s disease, achieving high-accuracy. We integrate Pleiades with leading protein biomarkers, achieving state-of-the-art results, underscoring the complementary value of epigenomic and proteomic multi-modal approaches. By advancing beyond the modelling of pure DNA sequences and relying on limited genomic regions, Pleiades establishes genome-wide epigenomic modelling as a new paradigm for clinical diagnostics, synthetic biology, and precision medicine. ### Competing Interest Statement Husam Babikir, Donal Byrne, Javkhlan-Ochir Ganbat, Anjeet Jhutty, Timing Liu, Hannah Madan, Christoforos Nalmpantis, Pouya Niki, Will Rowe, Ravi Solanki, Robert Sugar and Jonathan C. M. Wan are shareholders of Prima Mente. Henrik Zetterberg (HK) is a Wallenberg Scholar and a Distinguished Professor at the Swedish Research Council supported by grants from the Swedish Research Council (#2023-00356, #2022-01018 and #2019-02397), the European Union's Horizon Europe research and innovation programme under grant agreement No 101053962, and Swedish State Support for Clinical Research (#ALFGBG-71320). The UK DRI Biomarker Factory is funded by the National Institute for Health and Care Research University College London Hospitals Biomedical Research Centre, the UK Dementia Research Institute at UCL (UKDRI-1003), and the Weston Family Foundation. HZ has served at scientific advisory boards and/or as a consultant for Abbvie, Acumen, Alector, Alzinova, ALZpath, Amylyx, Annexon, Apellis, Artery Therapeutics, AZTherapies, Cognito Therapeutics, CogRx, Denali, Eisai, Enigma, LabCorp, Merck Sharp & Dohme, Merry Life, Nervgen, Novo Nordisk, Optoceutics, Passage Bio, Pinteon Therapeutics, Prothena, Quanterix, Red Abbey Labs, reMYND, Roche, Samumed, ScandiBio Therapeutics AB, Siemens Healthineers, Triplet Therapeutics, and Wave, has given lectures sponsored by Alzecure, BioArctic, Biogen, Cellectricon, Fujirebio, LabCorp, Lilly, Novo Nordisk, Oy Medix Biochemica AB, Roche, and WebMD, is a cofounder of Brain Biomarker Solutions in Gothenburg AB (BBS), which is a part of the GU Ventures Incubator Program, and is a shareholder of Prima Mente and MicThera (outside submitted work). Ivan Koychev (IK) has received honoraria for advisory board roles from J&J and Novo Nordisk and non-promotional speaker fees from Eisai, is in receipt of an investigatorinitiated grant from Novo Nordisk to explore the effects of a GLP-1 receptor agonist in preclinical dementia and is a medical advisor (stock options and/or retainer fees) to the following health technology companies: Five Lives, Oxford Brain Diagnostics, Leaf AI, Paloma Health and Prima Mente. Sofia Toniolo (ST), Masud Husain (MH), Sian Thompson (SiT) are funded by the Wellcome Trust. Sanjay G. Manohar (SGM) is funded by a Medical Research Council (MRC) Clinician Scientist Fellowship and National Institute of Health and Care Research (NIHR) Oxford Biomedical Research Centre (BRC) and NIHR Oxford Health BRC. MH has received speaker and advisory board honoraria from Lilly, Otsuka, and Sumitomo. Khaled Saab is a shareholder in Alphabet, Inc. Netanel Loyfer, ST, SiT, and SGM declare no conflict of interest. Two patents have been filed encompassing this work.

Apathy is a prevalent and persistent neuropsychiatric syndrome across many neurological disorders, significantly impacting both patients and caregivers. We systematically quantified discrepancies between self- and caregiver-reported apathy in 335 patients with a variety of diagnoses, such as frontotemporal dementia (behavioural variant and semantic dementia subtypes), Parkinson's disease, Parkinson's disease dementia, dementia with Lewy bodies, Alzheimer's disease dementia, mild cognitive impairment, small vessel cerebrovascular disease, subjective cognitive decline and autoimmune encephalitis. Using the Apathy Motivation Index (AMI) and its analogous caregiver version (AMI-CG), we found that caregiver-reported apathy consistently exceeded self-reported levels across all conditions. Moreover, self-reported apathy accounted for only 14.1% of the variance in caregiver ratings. This apathy reporting discrepancy was most pronounced in conditions associated with impaired insight, such as behavioural variant frontotemporal dementia, and was significantly correlated with cognitive impairment. Deficits in memory and fluency explained an additional 11.2% of the variance in caregiver-reported apathy. Specifically, executive function deficits (e.g. indexed by fluency) and memory impairments may contribute to behavioural inertia or recall of it. These findings highlight the need to integrate patient and caregiver perspectives in apathy assessments, especially for conditions with prominent cognitive impairment. To improve diagnostic accuracy and deepen our understanding of apathy across neurological disorders, we highlight the need for adapted apathy assessment strategies that account for cognitive impairment particularly in individuals with insight or memory deficits. Understanding the cognitive mechanisms underpinning discordant apathy reporting in dementia might help inform targeted clinical interventions and reduce caregiver burden.

Digital cognitive testing using online platforms has emerged as a potentially transformative tool in clinical neuroscience. In theory, it could provide a powerful means of screening for and tracking cognitive performance in people at risk of developing conditions such as Alzheimer's disease. Here we investigate whether digital metrics derived from an in-person administered, tablet-based short-term memory task-the 'What was where?' Oxford Memory Task-were able to clinically stratify patients at different points within the Alzheimer's disease continuum and to track disease progression over time. Performance of these metrics compared to traditional neuropsychological pen-and-paper screening tests of cognition was also analysed. A total of 325 people participated in this study: 49 patients with subjective cognitive decline, 57 with mild cognitive impairment, 63 with Alzheimer's disease dementia and 156 elderly healthy controls. Most digital metrics were able to discriminate between healthy controls and patients with mild cognitive impairment and between mild cognitive impairment and Alzheimer's disease patients. Some, including Absolute Localization Error, also differed significantly between patients with subjective cognitive decline and mild cognitive impairment. Identification accuracy was the best predictor of hippocampal atrophy, performing as well as standard screening neuropsychological tests. A linear support vector model combining digital metrics achieved high accuracy and performed at par with standard testing in discriminating between elderly healthy controls and subjective cognitive decline (area under the curve 0.82) and between subjective cognitive decline and mild cognitive impairment (area under the curve 0.92), while performing worse in classifying between mild cognitive impairment and Alzheimer's disease patients (area under the curve 0.75). Memory imprecision was able to predict cognitive decline on standard cognitive tests over one year. Overall, these findings show how it might be possible to use a digital memory test in clinics and clinical trial contexts to stratify and track performance across the Alzheimer's disease continuum. Toniolo et al. report that performance at a tablet-based short-term memory task, the 'What was where?' The Oxford Memory Task, was able to clinically stratify patients at different stages of Alzheimer's Disease, track disease progression longitudinally, predict cognitive decline on standard cognitive tests after 1 year and correlate to hippocampal atrophy.

Blood-based biomarkers enable early diagnosis of neurodegenerative diseases in a cost-effective and non-invasive way. Digital cognitive measures can capture early signs of cognitive impairment remotely and at scale. Their combination offers important opportunities for effective screening. We evaluated the performance of a new proteomic assay and a fully remote digital platform in a large memory clinic population (n = 391). Plasma biomarkers were measured using the NUcleic acid Linked Immuno-Sandwich Assay (NULISA) central nervous system panel. OCTAL (Oxford Cognitive Testing Portal) digital cognitive testing platform was used to measure cognition. Distinct proteomic patterns could be identified across different neurodegenerative diseases and associate with specific cognitive functions. Alzheimer's Disease (e.g., pTau217), and novel proteomics biomarkers (e.g., ACHE and IL6R) were highly correlated with worse cognition. A cluster of biomarkers, including PRDX6, was overexpressed in healthy controls and associated with better cognition, indexing cognitive resilience. ### Competing Interest Statement I.K. is a paid medical advisor for digital healthcare (Five Lives SAS, Lola Speaks) and biotechnology companies (Prima Mente, Oxford Brain Diagnostics) which have a focus on neurodegeneration. IK has received non-promotional speaker fees from Novo Nordisk and Eisai, and for advisory work for Johnson and Johnson, Zylorion and Novo Nordisk. He is in receipt of a grant for an investigator-initiated study from Novo Nordisk. S.T. has received speaker's honoraria for dedicated workshops from Eisai and Bial which are not related to this work. All other authors declare no financial or non-financial competing interests. ### Funding Statement This work was supported by the Wellcome Trust and the National Institute for Health Research (NIHR) Oxford Health Biomedical Research Centre. S.Z., S.T., M.B., A.S., A.F., M.H. are funded by the Wellcome Trust [226645/Z/22/Z]. S.T., S.G.M. and M.H. are also funded by the NIHR Oxford Health BRC. The project was further supported by a Guarantors of Brain post‐doctoral fellowship to S.T. S.G.M. is supported by the NIHR Oxford BRC. C.G. received a residency scholarship, funded by the Italian Ministry of University and Research (MUR), as part of the Neurology Residency Program at the Universita degli Studi di Milano. A.F. is further supported by funds from Medical Research Council [MR/X022013/1] and MyAware. CMvD is supported by the US National Insititute on Aging (NIH), NovoNordisk, the Oxford-GSK Institute of Molecular and Computational Medicine (IMCM), Centre of Artificial Intelligence for Precision Medicines (CAIPM) of the University of Oxford and King Abdul Aziz University, Alzheimer Research UK (ARUK), UK National Institute for Health and Care Research (NIHR) Oxford Biomedical Research Center (BRC), ZonMW (Delta Dementie) and Alzheimer Nederland. She is currently the Research Director Brain Health of the Health Data Research UK (HDR UK) and the UK Dementia Research Institute (UK DRI), working in partnership with Dementias Platform UK (DPUK). I.K. declares funding for this work through Alzheimer's Research UK, Alzheimer's Society and the People's Postcode Lottery READ OUT grant as well as the Medical Research Council (Dementias Platform UK) and the Oxford Health NIHR Biomedical Research Facility. The views expressed are those of the author(s) and not necessarily those of the funders. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The study was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments. Ethical approval was granted by the University of Oxford ethics committee (IRAS ID: 248379, Ethics Approval Reference: 18/SC/0448 and IRAS ID: 301319, Ethics Approval Reference: 22/WA/0183). All participants gave written informed consent prior to the start of the study. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes Data are available upon reasonable request. Access is contingent upon agreement with the original data providers (i.e. the researchers responsible for data collection) and subject to applicable ethical, legal, and confidentiality constraints.