Athanasios Anastasiou

Leaf aging of Amazonian canopy trees as revealed by spectral and physiochemical measurements.

Leaf aging is a fundamental driver of changes in leaf traits, thereby regulating ecosystem processes and remotely sensed canopy dynamics. We explore leaf reflectance as a tool to monitor leaf age and develop a spectra-based partial least squares regression (PLSR) model to predict age using data from a phenological study of 1099 leaves from 12 lowland Amazonian canopy trees in southern Peru. Results demonstrated monotonic decreases in leaf water (LWC) and phosphorus (Pmass ) contents and an increase in leaf mass per unit area (LMA) with age across trees; leaf nitrogen (Nmass ) and carbon (Cmass ) contents showed monotonic but tree-specific age responses. We observed large age-related variation in leaf spectra across trees. A spectra-based model was more accurate in predicting leaf age (R2 xa0=xa00.86; percent root mean square error (%RMSE)xa0=xa033) compared with trait-based models using single (R2 xa0=xa00.07-0.73; %RMSExa0=xa07-38) and multiple (R2 xa0=xa00.76; %RMSExa0=xa028) predictors. Spectra- and trait-based models established a physiochemical basis for the spectral age model. Vegetation indices (VIs) including the normalized difference vegetation index (NDVI), enhanced vegetation index 2 (EVI2), normalized difference water index (NDWI) and photosynthetic reflectance index (PRI) were all age-dependent. This study highlights the importance of leaf age as a mediator of leaf traits, provides evidence of age-related leaf reflectance changes that have important impacts on VIs used to monitor canopy dynamics and productivity and proposes a new approach to predicting and monitoring leaf age with important implications for remote sensing.

USING NHS PRIMARY CARE DATA TO IDENTIFY UNDIAGNOSED DEMENTIA

Background Up to 50% of patients with dementia may not receive a formal diagnosis, limiting access to appropriate services. It may be possible to build a picture of ‘underlying undiagnosed dementia’ from a profile of symptoms recorded in routine clinical practice. Aim To develop a machine learning tool to identify patients who may have underlying dementia but have not yet received formal diagnosis from analysis of routinely collected NHS data. Method Routinely collected NHS READ-encoded data were obtained from 18 consenting GP surgeries across Devon, UK, totalling 26,483 patient records of those aged >65 years. 539 Patients were identified as having dementia within the 2 year study period (June 2010 to June 2012). We determined other codes assigned to these patients that may contribute to dementia risk. The dataset was used to train a supervised classifier (Naives Bayes) to discriminate between patients with underlying dementia and healthy controls using a ten-fold cross-validation approach. Results The model obtained a sensitivity of 72.31% and a specificity of 83.06% for identifying dementia. Conclusion Routinely collected NHS data can be used to identify patients who are likely to have undiagnosed dementia. This type of methodology is promising for increasing dementia diagnosis within primary care.

Post-processing for spectral coherence of magnetoencephalogram background activity: Application to Alzheimer's disease

Estimating the connectivity between magnetoencephalogram (MEG) signals provides an excellent opportunity to analyze whole brain functional integration across a spectrum of conditions from health to disease. For this purpose, spectral coherence has been used widely as an easy-to-interpret metric of signal coupling. However, a number of systematic effects may influence the estimations of spectral coherence and subsequent inferences about brain activity. In this pilot study, we focus on the potentially confounding effects of the field spread and the on-going dynamic temporal variability inherent in the signals. We propose two simple post-processing approaches to account for these: 1) a jack-knife procedure to account for the variance in the estimation of spectral coherence; and 2) a detrending technique to reduce its dependence on sensor proximity. We illustrate the effect of these techniques in the estimation of MEG spectral coherence in the α band for 36 patients with Alzheimers disease and 26 control subjects.

Data modeling methods in clinical trials: experiences from the clinical trial methods in neurodegenerative diseases project

Objectives Clinical trials often generate large and diverse datasets. Data models are used to capture and organise the elements of the data in a meaningful way so that they can be stored and utilised by computer systems and support clinical decision making. This paper presents the data modeling considerations within the ‘Clinical Trial Methods in Neurodegenerative Diseases’ (CTMND) project funded by the NIHR [http://www.ctmnd.org]. The project adopts a holistic approach for the investigation of the suitability and efficiency of clinical observations in neurodegenerative diseases clinical studies. This ongoing research in novel clinical and surrogate outcome measures will be incorporated in an online data collection and analysis system to facilitate clinical trials and relevant research, taking into account, wherever possible, routinely collected NHS data. This paper presents ongoing research in the project’s data modeling aspects with the following objectives: 1. To review the current state of the art data models for capturing clinical information from the available literature. 2. To compare and contrast their features against the data management requirements of the project and outline the key factors that affected the adoption of a specific model for the CTMND project’s information system.

A novel thresholding method for the analysis of functional connectivity networks of the brain

Functional connectivity is a key concept in the analysis of brain function and it has been successfully applied to the study of the brain and changes caused by diseases such as Alzheimers Disease and Schizophrenia but also Aspergers syndrome. Thresholding is an essential operation in extracting functional connectivity networks and contributes to the determination of their structural characteristics. Several approaches to estimating the threshold value have been applied to functional modality datasets so far, but the choice of threshold value is largely adhoc. In this paper we propose a novel method for thresholding functional connectivity matrices that is universal, iterative and subject specific and evaluate its performance through simulations.

Novel metrics of functional network structure and their application to the detection and characterisation of Alzheimer’s disease

Metrics of network structure obtained using graph theory can provide valuable insights into the functioning and overall operation of neuronal networks in the brain. The objectives of this paper are to highlight the shortcomings of commonly used network structure metrics in the study of functional connectivity networks today and to propose a methodology for deriving more sensitive metrics which may be used to detect and characterize dementia and other disconnection syndromes.

Machine-learning based identification of undiagnosed dementia in primary care: a feasibility study

Background Up to half of patients with dementia may not receive a formal diagnosis, limiting access to appropriate services. It is hypothesised that it may be possible to identify undiagnosed dementia from a profile of symptoms recorded in routine clinical practice. Aim The aim of this study is to develop a machine learning-based model that could be used in general practice to detect dementia from routinely collected NHS data. The model would be a useful tool for identifying people who may be living with dementia but have not been formally diagnosed. Design & setting The study involved a case-control design and analysis of primary care data routinely collected over a 2-year period. Dementia diagnosed during the study period was compared to no diagnosis of dementia during the same period using pseudonymised routinely collected primary care clinical data. Method Routinely collected Read-encoded data were obtained from 18 consenting GP surgeries across Devon, for 26 483 patients aged >65 years. The authors determined Read codes assigned to patients that may contribute to dementia risk. These codes were used as features to train a machine-learning classification model to identify patients that may have underlying dementia. Results The model obtained sensitivity and specificity values of 84.47% and 86.67%, respectively. Conclusion The results show that routinely collected primary care data may be used to identify undiagnosed dementia. The methodology is promising and, if successfully developed and deployed, may help to increase dementia diagnosis in primary care.