Focus Ireland

Abstract BackgroundCurrent tools for Alzheimer's disease screening and staging used in clinical research (e.g. ACE‐3, ADAS‐Cog) require substantial face‐to‐face time with trained professionals, and may be affected by subjectivity, "white coat syndrome" and other biases. Alzheimer's symptomology is multi‐factorial and varies day‐to‐day. To enable stratification for precision treatments, more effective composite endpoints are required, with accurate quantification of functional impairment in individual domains. Ideally these would be measured objectively and frequently in real‐world contexts to improve validity, and to reduce clinical burden in trial and care settings.MethodWe present a machine‐learning stratification analysis using data from the Cumulus real‐world multi‐domain neuroassessment platform, in CNS101, a year‐long study designed with a consortium of 10 pharma companies. Participants completed a range of tablet‐based functional tasks with wake EEG, and separately recorded EEG during sleep. This cross‐sectional analysis uses 1111 sessions from repeated sampling with 101 participants (47 patients with mild Alzheimer’s type dementia, and 54 corresponding controls, recruited across 7 sites in the UK), during an early two‐week burst period of autonomous at‐home use. Bagging of decision trees was used to compare the power of different data sources in discriminating dementia from neurotypical states. This algorithm is widely used for its performance across learning tasks (linear and non‐linear), and heterogeneous datasets of differing size, levels of noise, and collinearity. For each set of input data, 100 classifiers were trained and evaluated (10‐fold cross‐validation in each of 10 random partitions), and the mean Area Under the Curve (AUC) of sensitivity/specificity trade‐off was calculated.ResultA full set of 691 multimodal features yielded a classifier with AUC performance of 0.953 and error rate 4.7% – compared to AUC of 0.955 and error rate 4.5% for the set of 13 ADAS‐Cog subscores. While individual modalities/domains all performed above chance, the 50 features from Cumulus cognitive tasks alone had an AUC performance of 0.951.ConclusionPatients using digital technology autonomously in the home can yield data that matches the discriminative power of a traditional clinician‐administered composite scale. This can enable objective precision measurement of disease severity at scale

IntroductionAccumulation of amyloid beta (Aβ), α‐synuclein (αSyn), and tau in dementias indicates their age‐related failure of elimination from the brain. Aβ is eliminated along basement membranes in walls of cerebral arterioles and leptomeningeal arteries (intramural peri‐arterial drainage [IPAD]); IPAD is impaired with age. We test the hypothesis that αSyn and tau are also eliminated from the normal brain along IPAD pathways.MethodsSoluble αSyn or tau was injected into mouse hippocampus. Animals were perfused 5 minutes to 7 days post‐injection. Blood vessels were identified by ROX‐SE for light‐sheet and immunolabeling for confocal microscopy. IPAD was quantified by measuring the proportion of arterioles with αSyn/tau.ResultsαSyn and tau are eliminated from the brain by IPAD but with different dynamics.DiscussionAge‐related failure of IPAD may play a role in the pathogenesis of synucleinopathies and tauopathies. αSyn persists within IPAD at 24 hours, which may affect immunotherapy for αSyn.

Estimates show that homelessness is increasing in Ireland. This study analysed the epidemiology of emergency hospitalisations among those experiencing homelessness between 2005-2014. All in-patient admissions to acute hospitals classified with 'no fixed abode' were extracted from the Hospital In-patient Enquiry System. Data were analysed using JMP. There were 2,051 in-patient emergency admissions of people classified with 'no fixed abode' during the study period, an increase of 406% since 2005 (78 in 2005 vs. 395 in 2014). The mean age was 40.6 (S.D. 13.2). The majority of patients (1,176 /2,051; 57%) had a mental/ behavioural diagnosis. Over one in ten (280; 13.7%) were admitted for ambulatory care sensitive conditions (ACSCs) including convulsions/epilepsy (N=92/280; 32.9%), cellulitis (62/280; 22.1%) and COPD (29/280; 10.4%). The health of homeless people is a fundamental issue that needs addressing. Access to, and use of, community and preventative services is needed to reduce utilisation of emergency hospital services.