Evaluation of Three State-of-the-Art Classifiers for Recognition of Activities of Daily Living from Smart Home Ambient Data

Smart homes for the aging population have recently started attracting the attention of the research community. The "health state" of smart homes is comprised of many different levels; starting with the physical health of citizens, it also includes longer-term health norms and outcomes, as well as the arena of positive behavior changes. One of the problems of interest is to monitor the activities of daily living (ADL) of the elderly, aiming at their protection and well-being. For this purpose, we installed passive infrared (PIR) sensors to detect motion in a specific area inside a smart apartment and used them to collect a set of ADL. In a novel approach, we describe a technology that allows the ground truth collected in one smart home to train activity recognition systems for other smart homes. We asked the users to label all instances of all ADL only once and subsequently applied data mining techniques to cluster in-home sensor firings. Each cluster would therefore represent the instances of the same activity. Once the clusters were associated to their corresponding activities, our system was able to recognize future activities. To improve the activity recognition accuracy, our system preprocessed raw sensor data by identifying overlapping activities. To evaluate the recognition performance from a 200-day dataset, we implemented three different active learning classification algorithms and compared their performance: naive Bayesian (NB), support vector machine (SVM) and random forest (RF). Based on our results, the RF classifier recognized activities with an average specificity of 96.53%, a sensitivity of 68.49%, a precision of 74.41% and an F-measure of 71.33%, outperforming both the NB and SVM classifiers. Further clustering markedly improved the results of the RF classifier. An activity recognition system based on PIR sensors in conjunction with a clustering classification approach was able to detect ADL from datasets collected from different homes. Thus, our PIR-based smart home technology could improve care and provide valuable information to better understand the functioning of our societies, as well as to inform both individual and collective action in a smart city scenario.

Application and comparison of classification algorithms for recognition of Alzheimer's disease in electrical brain activity (EEG)

The early detection of subjects with probable Alzheimer's disease (AD) is crucial for effective appliance of treatment strategies. Here we explored the ability of a multitude of linear and non-linear classification algorithms to discriminate between the electroencephalograms (EEGs) of patients with varying degree of AD and their age-matched control subjects. Absolute and relative spectral power, distribution of spectral power, and measures of spatial synchronization were calculated from recordings of resting eyes-closed continuous EEGs of 45 healthy controls, 116 patients with mild AD and 81 patients with moderate AD, recruited in two different centers (Stockholm, New York). The applied classification algorithms were: principal component linear discriminant analysis (PC LDA), partial least squares LDA (PLS LDA), principal component logistic regression (PC LR), partial least squares logistic regression (PLS LR), bagging, random forest, support vector machines (SVM) and feed-forward neural network. Based on 10-fold cross-validation runs it could be demonstrated that even tough modern computer-intensive classification algorithms such as random forests, SVM and neural networks show a slight superiority, more classical classification algorithms performed nearly equally well. Using random forests classification a considerable sensitivity of up to 85% and a specificity of 78%, respectively for the test of even only mild AD patients has been reached, whereas for the comparison of moderate AD vs. controls, using SVM and neural networks, values of 89% and 88% for sensitivity and specificity were achieved. Such a remarkable performance proves the value of these classification algorithms for clinical diagnostics.

Landscape connectivity, habitat structure and activity of bat guilds in farmland-dominated matrices

Agricultural intensification has caused a decline in structural elements in European farmland, where natural habitats are increasingly fragmented. The loss of habitat structures has a detrimental effect on biodiversity and affects bat species that depend on vegetation structures for foraging and commuting. We investigated the impact of connectivity and configuration of structural landscape elements on flight activity, species richness and diversity of insectivorous bats and distinguished three bat guilds according to species-specific bioacoustic characteristics. We tested whether bats with shorter-range echolocation were more sensitive to habitat fragmentation than bats with longer-range echolocation. We expected to find different connectivity thresholds for the three guilds and hypothesized that bats prefer linear over patchy landscape elements. Bat activity was quantified using repeated acoustic monitoring in 225 locations at 15 study plots distributed across the Swiss Central Plateau, where connectivity and the shape of landscape elements were determined by spatial analysis (GIS). Spectrograms of bat calls were assigned to species with the software batit by means of image recognition and statistical classification algorithms. Bat activity was significantly higher around landscape elements compared to open control areas. Short- and long-range echolocating bats were more active in well-connected landscapes, but optimal connectivity levels differed between the guilds. Species richness increased significantly with connectivity, while species diversity did not (Shannon's diversity index). Total bat activity was unaffected by the shape of landscape elements. Synthesis and applications. This study highlights the importance of connectivity in farmland landscapes for bats, with shorter-range echolocating bats being particularly sensitive to habitat fragmentation. More structurally diverse landscape elements are likely to reduce population declines of bats and could improve conditions for other declining species, including birds. Activity was highest around optimal values of connectivity, which must be evaluated for the different guilds and spatially targeted for a region's habitat configuration. In a multi-species approach, we recommend the reintroduction of structural elements to increase habitat heterogeneity should become part of agri-environment schemes.

Prestimulus default mode activity influences depth of processing and recognition in an emotional memory task

Low self-referential thoughts are associated with better concentration, which leads to deeper encoding and increases learning and subsequent retrieval. There is evidence that being engaged in externally rather than internally focused tasks is related to low neural activity in the default mode network (DMN) promoting open mind and the deep elaboration of new information. Thus, reduced DMN activity should lead to enhanced concentration, comprehensive stimulus evaluation including emotional categorization, deeper stimulus processing, and better long-term retention over one whole week. In this fMRI study, we investigated brain activation preceding and during incidental encoding of emotional pictures and on subsequent recognition performance. During fMRI, 24 subjects were exposed to 80 pictures of different emotional valence and subsequently asked to complete an online recognition task one week later. Results indicate that neural activity within the medial temporal lobes during encoding predicts subsequent memory performance. Moreover, a low activity of the default mode network preceding incidental encoding leads to slightly better recognition performance independent of the emotional perception of a picture. The findings indicate that the suppression of internally-oriented thoughts leads to a more comprehensive and thorough evaluation of a stimulus and its emotional valence. Reduced activation of the DMN prior to stimulus onset is associated with deeper encoding and enhanced consolidation and retrieval performance even one week later. Even small prestimulus lapses of attention influence consolidation and subsequent recognition performance. Hum Brain Mapp, 2015. © 2015 Wiley Periodicals, Inc.

Role of complement and Fc{gamma} receptors in the protective activity of the long pentraxin PTX3 against Aspergillus fumigatus

Pentraxin 3 (PTX3) is a soluble pattern recognition molecule playing a nonredundant role in resistance against Aspergillus fumigatus. The present study was designed to investigate the molecular pathways involved in the opsonic activity of PTX3. The PTX3 N-terminal domain was responsible for conidia recognition, but the full-length molecule was necessary for opsonic activity. The PTX3-dependent pathway of enhanced neutrophil phagocytic activity involved complement activation via the alternative pathway; Fc receptor (Fc R) IIA/CD32 recognition of PTX3-sensitized conidia and complement receptor 3 (CR3) activation; and CR3 and CD32 localization to the phagocytic cup. Gene targeted mice (ptx3, FcR common chain, C3, C1q) validated the in vivo relevance of the pathway. In particular, the protective activity of exogenous PTX3 against A fumigatus was abolished in FcR common chain-deficient mice. Thus, the opsonic and antifungal activity of PTX3 is at the crossroad between complement, complement receptor 3-, and Fc R-mediated recognition. Because short pentraxins (eg, C-reactive protein) interact with complement and Fc R, the present results may have general significance for the mode of action of these components of the humoral arm of innate immunity.

Socio-Spatial Relationships in Dairy Cows

Farm animals may serve as models for evaluating social networks in a controlled environment. We used an automated system to track, at fine temporal and spatial resolution (once per minute, +/- 50 cm) every individual in six herds of dairy cows (Bos taurus). We then analysed the data using social network analyses. Relationships were based on non-random attachment and avoidance relationships in respect to synchronous use and distances observed in three different functional areas (activity, feeding and lying). We found that neither synchrony nor distance between cows was strongly predictable among the three functional areas. The emerging social networks were tightly knit for attachment relationships and less dense for avoidance relationships. These networks loosened up from the feeding and lying area to the activity area, and were less dense for relationships based on synchronicity than on median distance with respect to node degree, relative size of the largest cluster, density and diameter of the network. In addition, synchronicity was higher in dyads of dairy cows that had grown up together and shared their last dry period. This last effect disappeared with increasing herd size. Dairy herds can be characterized by one strongly clustered network including most of the herd members with many non-random attachment and avoidance relationships. Closely synchronous dyads were composed of cows with more intense previous contact. The automatic tracking of a large number of individuals proved promising in acquiring the data necessary for tackling social network analyses.

The Ras inhibitors caveolin-1 and docking protein 1 activate peroxisome proliferator-activated receptor ? through spatial relocalization at helix 7 of its ligand-binding domain

Peroxisome proliferator-activated receptor ? (PPAR?) is a transcription factor that promotes differentiation and cell survival in the stomach. PPAR? upregulates and interacts with caveolin-1 (Cav1), a scaffold protein of Ras/mitogen-activated protein kinases (MAPKs). The cytoplasmic-to-nuclear localization of PPAR? is altered in gastric cancer (GC) patients, suggesting a so-far-unknown role for Cav1 in spatial regulation of PPAR? signaling. We show here that loss of Cav1 accelerated proliferation of normal stomach and GC cells in vitro and in vivo. Downregulation of Cav1 increased Ras/MAPK-dependent phosphorylation of serine 84 in PPAR? and enhanced nuclear translocation and ligand-independent transcription of PPAR? target genes. In contrast, Cav1 overexpression sequestered PPAR? in the cytosol through interaction of the Cav1 scaffolding domain (CSD) with a conserved hydrophobic motif in helix 7 of PPAR?'s ligand-binding domain. Cav1 cooperated with the endogenous Ras/MAPK inhibitor docking protein 1 (Dok1) to promote the ligand-dependent transcriptional activity of PPAR? and to inhibit cell proliferation. Ligand-activated PPAR? also reduced tumor growth and upregulated the Ras/MAPK inhibitors Cav1 and Dok1 in a murine model of GC. These results suggest a novel mechanism of PPAR? regulation by which Ras/MAPK inhibitors act as scaffold proteins that sequester and sensitize PPAR? to ligands, limiting proliferation of gastric epithelial cells.

Mutagenesis and computer modeling studies of a GPCR conserved residue W5.43(194) in ligand recognition and signal transduction for CB2 receptor

W5.43(194), a conserved tryptophan residue among G-protein coupled receptors (GPCRs) and cannabinoid receptors (CB), was examined in the present report for its significance in CB2 receptor ligand binding and adenylyl cyclase (AC) activity. Computer modeling postulates that this site in CB2 may be involved in the affinity of WIN55212-2 and SR144528 through aromatic contacts. In the present study, we reported that a CB2 receptor mutant, W5.43(194)Y, which had a tyrosine (Y) substitution for tryptophan (W), retained the binding affinity for CB agonist CP55940, but reduced binding affinity for CB2 agonist WIN55212-2 and inverse agonist SR144528 by 8-fold and 5-fold, respectively; the CB2 W5.43(194)F and W5.43(194)A mutations significantly affect the binding activities of CP55940, WIN55212-2 and SR144528. Furthermore, we found that agonist-mediated inhibition of the forskolin-induced cAMP production was dramatically diminished in the CB2 mutant W5.43(194)Y, whereas W5.43(194)F and W5.43(194)A mutants resulted in complete elimination of downstream signaling, suggesting that W5.43(194) was essential for the full activation of CB2. These results indicate that both aromatic interaction and hydrogen bonding are involved in ligand binding for the residue W5.43(194), and the mutations of this tryptophan site may affect the conformation of the ligand binding pocket and therefore control the active conformation of the wild type CB2 receptor. W5.43(194)Y/F/A mutations also displayed noticeable enhancement of the constitutive activation probably attributed to the receptor conformational changes resulted from the mutations.

Callosal connections of primary visual cortex predict the spatial spreading of binocular rivalry across the visual hemifields

In binocular rivalry, presentation of different images to the separate eyes leads to conscious perception alternating between the two possible interpretations every few seconds. During perceptual transitions, a stimulus emerging into dominance can spread in a wave-like manner across the visual field. These traveling waves of rivalry dominance have been successfully related to the cortical magnification properties and functional activity of early visual areas, including the primary visual cortex (V1). Curiously however, these traveling waves undergo a delay when passing from one hemifield to another. In the current study, we used diffusion tensor imaging (DTI) to investigate whether the strength of interhemispheric connections between the left and right visual cortex might be related to the delay of traveling waves across hemifields. We measured the delay in traveling wave times (ΔTWT) in 19 participants and repeated this test 6 weeks later to evaluate the reliability of our behavioral measures. We found large interindividual variability but also good test-retest reliability for individual measures of ΔTWT. Using DTI in connection with fiber tractography, we identified parts of the corpus callosum connecting functionally defined visual areas V1-V3. We found that individual differences in ΔTWT was reliably predicted by the diffusion properties of transcallosal fibers connecting left and right V1, but observed no such effect for neighboring transcallosal visual fibers connecting V2 and V3. Our results demonstrate that the anatomical characteristics of topographically specific transcallosal connections predict the individual delay of interhemispheric traveling waves, providing further evidence that V1 is an important site for neural processes underlying binocular rivalry.

Debris-flow activity in abandoned channels of the Manival torrent reconstructed with LiDAR and tree-ring data

Hydrogeomorphic processes are a major threat in many parts of the Alps, where they periodically damage infrastructure, disrupt transportation corridors or even cause loss of life. Nonetheless, past torrential activity and the analysis of areas affected during particular events remain often imprecise. It was therefore the purpose of this study to reconstruct spatio-temporal patterns of past debris-flow activity in abandoned channels on the forested cone of the Manival torrent (Massif de la Chartreuse, French Prealps). A Light Detecting and Ranging (LiDAR) generated Digital Elevation Model (DEM) was used to identify five abandoned channels and related depositional forms (lobes, lateral levees) in the proximal alluvial fan of the torrent. A total of 156 Scots pine trees (Pinus sylvestris L.) with clear signs of debris flow events was analyzed and growth disturbances (GD) assessed, such as callus tissue, the onset of compression wood or abrupt growth suppression. In total, 375 GD were identified in the tree-ring samples, pointing to 13 debris-flow events for the period 1931–2008. While debris flows appear to be very common at Manival, they have only rarely propagated outside the main channel over the past 80 years. Furthermore, analysis of the spatial distribution of disturbed trees contributed to the identification of four patterns of debris-flow routing and led to the determination of three preferential breakout locations. Finally, the results of this study demonstrate that the temporal distribution of debris flows did not exhibit significant variations since the beginning of the 20th century.

Detection of gelatinolytic activity in developing basement membranes of the mouse embryo head by combining sensitive in situ zymography with immunolabeling

Genetic evidence indicates that the major gelatinases MMP-2 and MMP-9 are involved in mammalian craniofacial development. Since these matrix metalloproteinases are secreted as proenzymes that require activation, their tissue distribution does not necessarily reflect the sites of enzymatic activity. Information regarding the spatial and temporal expression of gelatinolytic activity in the head of the mammalian embryo is sparse. Sensitive in situ zymography with dye-quenched gelatin (DQ-gelatin) has been introduced recently; gelatinolytic activity results in a local increase in fluorescence. Using frontal sections of wild-type mouse embryo heads from embryonic day 14.5-15.5, we optimized and validated a simple double-labeling in situ technique for combining DQ-gelatin zymography with immunofluorescence staining. MMP inhibitors were tested to confirm the specificity of the reaction in situ, and results were compared to standard SDS-gel zymography of tissue extracts. Double-labeling was used to show the spatial relationship in situ between gelatinolytic activity and immunostaining for gelatinases MMP-2 and MMP-9, collagenase 3 (MMP-13) and MT1-MMP (MMP-14), a major activator of pro-gelatinases. Strong gelatinolytic activity, which partially overlapped with MMP proteins, was confirmed for Meckel's cartilage and developing mandibular bone. In addition, we combined in situ zymography with immunostaining for extracellular matrix proteins that are potential gelatinase substrates. Interestingly, gelatinolytic activity colocalized precisely with laminin-positive basement membranes at specific sites around growing epithelia in the developing mouse head, such as the ducts of salivary glands or the epithelial fold between tongue and lower jaw region. Thus, this sensitive method allows to associate, with high spatial resolution, gelatinolytic activity with epithelial morphogenesis in the embryo.

Altered cortico-basal ganglia motor pathways reflect reduced volitional motor activity in schizophrenia

Little is known about the neurobiology of hypokinesia in schizophrenia. Therefore, the aim of this study was to investigate alterations of white matter motor pathways in schizophrenia and to relate our findings to objectively measured motor activity. We examined 21 schizophrenia patients and 21 healthy controls using diffusion tensor imaging and actigraphy. We applied a probabilistic fibre tracking approach to investigate pathways connecting the dorsolateral prefrontal cortex (dlPFC), the rostral anterior cingulate cortex (rACC), the pre-supplementary motor area (pre-SMA), the supplementary motor area proper (SMA-proper), the primary motor cortex (M1), the caudate nucleus, the striatum, the pallidum and the thalamus. Schizophrenia patients had lower activity levels than controls. In schizophrenia we found higher probability indices forming part of a bundle of interest (PIBI) in pathways connecting rACC, pre-SMA and SMA-proper as well as in pathways connecting M1 and pre-SMA with caudate nucleus, putamen, pallidum and thalamus and a reduced spatial extension of motor pathways in schizophrenia. There was a positive correlation between PIBI and activity level in the right pre-SMA-pallidum and the left M1-thalamus connection in healthy controls, and in the left pre-SMA-SMA-proper pathway in schizophrenia. Our results point to reduced volitional motor activity and altered motor pathway organisation in schizophrenia. The identified associations between the amount of movement and structural connectivity of motor pathways suggest dysfunction of cortico-basal ganglia pathways in the pathophysiology of hypokinesia in schizophrenia. Schizophrenia patients may use cortical pathways involving the supplementary motor area to compensate for basal ganglia dysfunction.