Sliding-Window Normalization to Improve the Performance of Machine-Learning Models for Real-Time Motion Prediction Using Electromyography.

Many researchers have used machine learning models to control artificial hands, walking aids, assistance suits, etc., using the biological signal of electromyography (EMG). The use of such devices requires high classification accuracy. One method for improving the classification performance of machine learning models is normalization, such as z-score. However, normalization is not used in most EMG-based motion prediction studies because of the need for calibration and fluctuation of reference value for calibration (cannot re-use). Therefore, in this study, we proposed a normalization method that combines sliding-window and z-score normalization that can be implemented in real-time processing without need for calibration. The effectiveness of this normalization method was confirmed by conducting a single-joint movement experiment of the elbow and predicting its rest, flexion, and extension movements from the EMG signal. The proposed method achieved 77.7% accuracy, an improvement of 21.5% compared to the non-normalization (56.2%). Furthermore, when using a model trained by other people's data for application without calibration, the proposed method achieved 63.1% accuracy, an improvement of 8.8% compared to the z-score (54.4%). These results showed the effectiveness of the simple and easy-to-implement method, and that the classification performance of the machine learning model could be improved.

[Promoting community gathering places "Kayoinoba" for healthy aging reduce health inequalities among communities: An eight-year ecological study].

Objectives　This study aimed to investigate whether health inequalities among communities would be reduced by intensively enhancing the "Kayoinoba" program in model communities where many high-risk, older adults live.Methods　Kobe City and the Japan Gerontological Evaluation Study created a mail survey for older adults in 78 communities (community ≈ junior high school district) to conduct community diagnosis. Sixteen communities showed poor values along multiple dimensions of risk and required priority measures. From 2014 to 2019, we designated these 16 communities as model communities. Then, municipal officials and researchers cooperated to support the establishment and management of "Kayoinoba." By using four-waves of mail survey data (in 2011, 2013, 2016, and 2019 with n=8,872, 10,572, 10,063, and 5,759, respectively), secular transitions of nine intermediate outcome indicators (three=social participation, two=social network, and four=social support) and five health outcome indicators (physical function, malnutrition, oral function, cognitive function, and depressive symptoms) were compared between model (n=16) and non-model (n=62) communities via multilevel mixed-effects linear regression analysis.Results　In the 2011 and 2013 surveys, model communities showed poor value compared to the non-model communities in 13 of the 14 indicators. A significant interaction between the year and model/non-model communities was confirmed for four intermediate outcome indicators (sports and hobby group participation, number of friends met, and providing emotional support) and three health outcome indicators (oral function, cognitive function, and depressive symptoms). The differences were reduced or eliminated in the 2016 and 2019 surveys. For example, hobby group participation in 2011 was 29.7% vs. 35.0% in model vs. non-model communities; the difference narrowed to 35.2% vs. 36.1% (P=0.008). Similarly, providing emotional support increased from 83.9% vs. 87.0% to 93.3% vs. 93.3% (P=0.007). Depressive symptoms decreased from 31.4% vs. 27.2% to 18.6% vs. 20.3% (P<0.001).Conclusions　Promoting community gathering places "Kayoinoba" for six years in communities where many high-risk older adults live may foster social participation, networking, and support and may help reduce health inequalities among communities.

An auditory brain-computer interface to detect changes in sound pressure level for automatic volume control.

Volume control is necessary to adjust sound levels for a comfortable audio or video listening experience. This study aims to develop an automatic volume control system based on a brain-computer interface (BCI). We thus focused on a BCI using an auditory oddball paradigm, and conducted two types of experiments. In the first experiment, the participant was asked to pay attention to a target sound where the sound level was high (70 dB) compared with the other sounds (60 dB). The brain activity measured by electroencephalogram showed large positive activity (P300) for the target sound, and classification of the target and nontarget sounds achieved an accuracy of 0.90. The second experiment adopted a two-target paradigm where a low sound level (50 dB) was introduced as the second target sound. P300 was also observed in the second experiment, and a value of 0.76 was obtained for the binary classification of the target and nontarget sounds. Further, we found that better accuracy was observed in large sound levels compared to small ones. These results suggest the possibility of using BCI for automatic volume control; however, it is necessary to improve its accuracy for application in daily life.

Diversity, heterogeneity and orientation-dependent variation of spike count correlation in the cat visual cortex.

Cortical neurons are known to be noisy encoders of information, showing large response variabilities with repeated presentations of identical stimuli. These spike count variabilities are correlated over the cell population and their neuronal mechanism and functional significance have not been well understood. Recently there has been much debate over the magnitude of the population mean of the correlation, ranging from 0.1 to 0.2 down to nearly zero. We performed multi-neuron recordings on the cat visual cortex and found that the population mean did not necessarily represent the nature of correlated variabilities because the spike count correlation showed significant diversity and heterogeneity. Although the population mean was relatively small (0.06), the correlations of individual unit pairs were distributed over a broad range, extending to both positive and negative values. In most of the recording sessions of local cell populations (83%), significantly positive correlations coexisted with significantly negative ones in different unit pairs. Furthermore, nearly 20% of the unit pairs showed significant variation in the spike count correlation for different stimulus orientations. Correlation analysis between the spike count correlation and the firing activity of the unit pair suggested that the orientation tuning properties of the two quantities were unlikely to have originated from a common neuronal mechanism. Diversity, heterogeneity and context-dependent variation suggests that the correlated spike count variabilities originate not from fixed anatomical connections but rather from the dynamic interaction of neuronal networks.

Verification of Normalization Method to Improve Usability and Versatility among Users of Applications that Predict Continuous Motion Using Electromyography.

In applications using electromyography (EMG), it is important to ensure high performance for all users (versatility among users) and to enable use without prior preparation (usability). Some of the current applications that use EMG normalize the signal through methods based on the measured maximum absolute value of EMG (maEMG), such as dynamic contraction (DC). However, usability is low when using DC because the reference value must be measured first every time the application is used. Further, the versatility among users is low because of the nonlinearity of EMG and the fact that maEMG varies among users. This study aimed to improve usability and versatility among users for continuous classification tasks using EMG. To this end, we developed a normalization method using sliding-window and z-score normalization techniques. The results reveal that the proposed method exhibits higher usability and versatility among users than DC. The proposed method does not require any calibration time, suggesting improved usability, while yielding the same classification accuracy as DC (57% for three target tasks) for a model trained using a subject's own data. Further, for a model trained with other users' data, the proposed method yields a classification accuracy of 53%, which is 18% higher than that of DC (35%), suggesting versatility among users. These results demonstrate that the proposed normalization method improves usability and versatility for users of practical applications that use EMG and perform continuous classification, such as prosthetic hands.

Assessing the Impacts of Correlated Variability with Dissociated Timescales.

Despite the profound influence on coding capacity of sensory neurons, the measurements of noise correlations have been inconsistent. This is, possibly, because nonstationarity, i.e., drifting baselines, engendered the spurious long-term correlations even if no actual short-term correlation existed. Although attempts to separate them have been made previously, they were ad hoc for specific cases or computationally too demanding. Here we proposed an information-geometric method to unbiasedly estimate pure short-term noise correlations irrespective of the background brain activities without demanding computational resources. First, the benchmark simulations demonstrated that the proposed estimator is more accurate and computationally efficient than the conventional correlograms and the residual correlations with Kalman filters or moving averages of length three or more, while the best moving average of length two coincided with the propose method regarding correlation estimates. Next, we analyzed the cat V1 neural responses to demonstrate that the statistical test accompanying the proposed method combined with the existing nonstationarity test enabled us to dissociate short-term and long-term noise correlations. When we excluded the spurious noise correlations of purely long-term nature, only a small fraction of neuron pairs showed significant short-term correlations, possibly reconciling the previous inconsistent observations on existence of significant noise correlations. The decoding accuracy was slightly improved by the short-term correlations. Although the long-term correlations deteriorated the generalizability, the generalizability was recovered by the decoder with trend removal, suggesting that brains could overcome nonstationarity. Thus, the proposed method enables us to elucidate the impacts of short-term and long-term noise correlations in a dissociated manner.

Design of multielectrode arrays for uniform sampling of different orientations of tuned unit populations in the cat visual cortex.

For better reconstruction of stimulus orientation from a single trial activity of the neuron population in the visual cortex, we need uniform samplings of differently oriented tuned neurons. We recorded multiple neurons simultaneously by using either a four-tetrode array or an eight-microelectrode array, and examined what kinds of electrodes and layouts provided a more homogeneous distribution of the units' optimal orientations. The unit population sampled by a four-tetrode array showed more homogeneous distribution than those sampled by an eight-microelectrode array. We confirmed this property by simulated recording sessions based on the optical imaging data of the orientation map.

Galpha(12/13) mediates alpha(1)-adrenergic receptor-induced cardiac hypertrophy.

In neonatal cardiomyocytes, activation of the G(q)-coupled alpha(1)-adrenergic receptor (alpha(1)AR) induces hypertrophy by activating mitogen-activated protein kinases, including c-Jun NH(2)-terminal kinase (JNK). Here, we show that JNK activation is essential for alpha(1)AR-induced hypertrophy, in that alpha(1)AR-induced hypertrophic responses, such as reorganization of the actin cytoskeleton and increased protein synthesis, could be blocked by expressing the JNK-binding domain of JNK-interacting protein-1, a specific inhibitor of JNK. We also identified the classes and subunits of G proteins that mediate alpha(1)AR-induced JNK activation and hypertrophic responses by generating several recombinant adenoviruses that express polypeptides capable of inhibiting the function of specific G-protein subunits. alpha(1)AR-induced JNK activation was inhibited by the expression of carboxyl terminal regions of Galpha(q), Galpha(12), and Galpha(13). JNK activation was also inhibited by the Galpha(q/11)- or Galpha(12/13)-specific regulator of G-protein signaling (RGS) domains and by C3 toxin but was not affected by treatment with pertussis toxin or by expression of the carboxyl terminal region of G protein-coupled receptor kinase 2, a polypeptide that sequesters Gbetagamma. alpha(1)AR-induced hypertrophic responses were inhibited by Galpha(q/11)- and Galpha(12/13)-specific RGS domains, C3 toxin, and the carboxyl terminal region of G protein-coupled receptor kinase 2 but not by pertussis toxin. Activation of Rho was inhibited by carboxyl terminal regions of Galpha(12) and Galpha(13) but not by Galpha(q). Our findings suggest that alpha(1)AR-induced hypertrophic responses are mediated in part by a Galpha(12/13)-Rho-JNK pathway, in part by a G(q/11)-JNK pathway that is Rho independent, and in part by a Gbetagamma pathway that is JNK independent.

Heterotrimeric G protein G alpha13-induced induction of cytokine mRNAs through two distinct pathways in cardiac fibroblasts.

Overexpression of constitutively active (CA)-G alpha13 significantly increased the expression of interleukin (IL)-1beta and IL-6 mRNAs and proteins in rat cardiac fibroblasts. IL-1beta mRNA induction by CA-G alpha13 was suppressed by diphenyleneiodonium (DPI), an NADPH oxidase inhibitor, but not by BAPTA-AM, an intracellular Ca2+ chelator. In contrast, IL-6 mRNA induction by CA-G alpha13 was suppressed by BAPTA-AM but not by DPI. However, both IL-1beta and IL-6 mRNA induction was suppressed by nuclear factor kappaB (NF-kappaB) inhibitors. The CA-G alpha13-induced NF-kappaB activation was suppressed by DPI and BAPTA-AM, but not C3 toxin and the Rho-kinase inhibitor Y27632. IL-6 mRNA induction by CA-G alpha13 was suppressed by SK&F96365 (1-[beta-[3-(4-methoxyphenyl)propoxy]-4-methoxyphenethyl]-1H-imidazole hydrochloride), an inhibitor of receptor-activated nonselective cation channels, and the expression of CA-G alpha13 increased basal Ca2+ influx. These results suggest that G alpha13 regulates IL-1beta mRNA induction through the reactive oxygen species-NF-kappaB pathway, while it regulates IL-6 mRNA induction through the Ca2+-NF-kappaB pathway.

Caveolae-independent activation of protein kinase A in rat neonatal myocytes.

Cardiomyocytes express both beta(1)- and beta(2)-adrenergic receptors, and these receptors play a differential role in chronotropic and inotropic effects of the heart. Caveolae are known as an important regulator of G-protein-coupled receptor signaling. In the present report, we examined whether caveolae have a role in beta-adrenergic receptor-stimulated cAMP production and protein kinase A activation in neonatal myocytes. Isoproterenol-stimulated cAMP production was mediated by beta(1)- and beta(2)-subtypes, which depends on the receptor number of each subtype. However, protein kinase A activation was exclusively mediated by the beta(1)-subtype. Disruption of caveolae by methyl-beta-cyclodextrin treatment did not affect the relative contribution of subtypes to isoproterenol-stimulated cAMP production. beta(1)-Subtype-mediated protein kinase A activation was also not affected by the disruption of caveolae. These results suggest that beta(1)-adrenergic receptor-mediated protein kinase A activation is compartmentalized and independent of caveolae.

G alpha 12/13- and reactive oxygen species-dependent activation of c-Jun NH2-terminal kinase and p38 mitogen-activated protein kinase by angiotensin receptor stimulation in rat neonatal cardiomyocytes.

In the present study, we examined signal transduction mechanism of reactive oxygen species (ROS) production and the role of ROS in angiotensin II-induced activation of mitogen-activated protein kinases (MAPKs) in rat neonatal cardiomyocytes. Among three MAPKs, c-Jun NH(2)-terminal kinase (JNK) and p38 MAPK required ROS production for activation, as an NADPH oxidase inhibitor, diphenyleneiodonium, inhibited the activation. The angiotensin II-induced activation of JNK and p38 MAPK was also inhibited by the expression of the Galpha(12/13)-specific regulator of G protein signaling (RGS) domain, a specific inhibitor of Galpha(12/13), but not by an RGS domain specific for Galpha(q). Constitutively active Galpha(12)- or Galpha(13)-induced activation of JNK and p38 MAPK, but not extracellular signal-regulated kinase (ERK), was inhibited by diphenyleneiodonium. Angiotensin II receptor stimulation rapidly activated Galpha(13), which was completely inhibited by the Galpha(12/13)-specific RGS domain. Furthermore, the Galpha(12/13)-specific but not the Galpha(q)-specific RGS domain inhibited angiotensin II-induced ROS production. Dominant negative Rac inhibited angiotensin II-stimulated ROS production, JNK activation, and p38 MAPK activation but did not affect ERK activation. Rac activation was mediated by Rho and Rho kinase, because Rac activation was inhibited by C3 toxin and a Rho kinase inhibitor, Y27632. Furthermore, angiotensin II-induced Rho activation was inhibited by Galpha(12/13)-specific RGS domain but not dominant negative Rac. An inhibitor of epidermal growth factor receptor kinase AG1478 did not affect angiotensin II-induced JNK activation cascade. These results suggest that Galpha(12/13)-mediated ROS production through Rho and Rac is essential for JNK and p38 MAPK activation.

Galpha12/13-mediated production of reactive oxygen species is critical for angiotensin receptor-induced NFAT activation in cardiac fibroblasts.

Angiotensin II (Ang II) activates multiple signaling pathways leading to hyperplasia of cardiac fibroblasts. Reactive oxygen species (ROS) produced by Ang II stimulation are assumed to play pivotal roles in this process. Here, we show that ROS mediate Ang II-induced activation of nuclear factor of activated T cells (NFAT) in rat cardiac fibroblasts. Ang II-induced NFAT activation was suppressed by diphenyleneiodonium (an NADPH oxidase inhibitor), dominant negative (DN)-Rac, DN-p47(phox), and an inhibitor of Galpha(12/13) (Galpha(12/13)-specific regulator of G protein signaling domain of p115RhoGEF, p115-regulator of G protein signaling (RGS)). Stimulation of Ang II receptor increased the intracellular ROS level in a Rac- and p47(phox)-dependent manner. Because p115-RGS suppressed Ang II-induced Rac activation, Ang II receptor-coupled Galpha(12/13) mediated NFAT activation through ROS production by Rac activation. Ang II-induced nuclear translocation of the green fluorescent protein (GFP)-tagged amino-terminal region of NFAT4 (GFP-NFAT4) was suppressed by p115-RGS or BAPTA but not by diphenyleneiodonium. The expression of constitutively active (CA)-Galpha(12/13), CA-G translocation alpha(13), or CA-Rac increased the nuclear of GFP-NFAT4. These results suggest that NFAT activity is regulated by both Ca(2+)-dependent and ROS-dependent pathways. Furthermore, activation of c-Jun NH(2)-terminal kinase (JNK) induced by Ang II stimulation is required for NFAT activation because Ang II-induced NFAT activation was inhibited by SP600125, a selective JNK inhibitor. These results indicate that Ang II stimulates the nuclear translocation and activation of NFAT by integrated pathways including the activation of Galpha(12/13), Rac, NADPH oxidase, and JNK and that Galpha(12/13)-mediated ROS production is essential for NFAT transcriptional activation.

G beta gamma counteracts G alpha(q) signaling upon alpha(1)-adrenergic receptor stimulation.

In rat neonatal myocytes, a constitutively active G alpha(q) causes cellular injury and apoptosis. However, stimulation of the alpha(1)-adrenergic receptor, one of the G(q) protein-coupled receptors, with phenylephrine for 48 h causes little cellular injury and apoptosis. Expression of the G beta gamma-sequestering peptide beta ARK-ct increases the phenylephrine-induced cardiac injury, indicating that G beta gamma released from G(q) counteracts the G alpha(q)-mediated cellular injury. Stimulation with phenylephrine activates extracellular signal-regulated kinase (ERK) and Akt, and activation is significantly blunted by beta ARK-ct. Inhibition of Akt by inhibitors of phosphatidylinositol 3-kinase increases the cellular injury induced by phenylephrine stimulation. In contrast to the inhibition of Akt, inhibition of ERK does not affect the phenylephrine-induced cardiac injury. These results suggest that G beta gamma released from G(q) upon alpha(1)-adrenergic receptor stimulation activates ERK and Akt. However, activation of Akt but not ERK plays an important role in the protection against the G alpha(q)-induced cellular injury and apoptosis.

Differential requirement of G alpha12, G alpha13, G alphaq, and G beta gamma for endothelin-1-induced c-Jun NH2-terminal kinase and extracellular signal-regulated kinase activation.

In the present study, we examined the roles of G(12), G(13), G(q), and G(i) in endothelin-1-induced hypertrophic responses. Endothelin-1 stimulation activated extracellular signal-regulated kinase (ERK) and c-Jun NH(2)-terminal kinase (JNK) in cultured rat neonatal myocytes. The activation of JNK, but not ERK, was inhibited by the expression of carboxyl terminal regions of G alpha(12) and G alpha(13). JNK activation was also inhibited by expression of the G alpha(12)/G alpha(13)-specific inhibitor regulator of G protein signaling (RGS) domain of p115RhoGEF and the G alpha(q)-specific inhibitor RGS domain of the G protein-coupled receptor kinase 2 (GRK2-RGS). JNK activation was not, however, inhibited by expression of the carboxyl terminal region of G protein-coupled receptor kinase 2 (GRK2-ct), which is a G beta gamma-sequestering polypeptide. Additionally, JNK activation but not ERK activation was inhibited by the expression of C3 exoenzyme that inactivates small GTPase Rho. These results suggest that JNK activation by G alpha(12), G alpha(13), and G alpha(q) is involved in Rho. On the other hand, ERK activation was inhibited by pertussis toxin treatment, the receptor-G(i) uncoupler, and GRK2-ct. Thus, ERK was activated by G alpha(i)- and G beta gamma-dependent pathways. These results clearly demonstrate that differential pathways activate JNK and ERK.