Search for: All records

Fall prevention has always been a crucial topic for injury prevention. Research shows that real-time posture monitoring and subsequent fall prevention are important for the prevention of fall-related injuries. In this research, we determine a real-time posture classifier by comparing classical and deep machine learning classifiers in terms of their accuracy and robustness for posture classification. For this, multiple classical classifiers, including classical machine learning, support vector machine, random forest, neural network, and Adaboost methods, were used. Deep learning methods, including LSTM and transformer, were used for posture classification. In the experiment, joint data were obtained using an RGBD camera. The results show that classical machine learning posture classifier accuracy was between 75% and 99%, demonstrating that the use of classical machine learning classification alone is sufficient for real-time posture classification even with missing joints or added noise. The deep learning method LSTM was also effective in classifying the postures with high accuracy, despite incurring a significant computational overhead cost, thus compromising the real-time posture classification performance. The research thus shows that classical machine learning methods are worthy of our attention, at least, to consider for reuse or reinvention, especially for real-time posture classification tasks. The insight of using a classical posture classifier for large-scale human posture classification is also given through this research. 

In contrast to their spontaneous deprotonation in aqueous solution, reactions of guanine and guanosine radical cations with water in the gas phase are exclusively initiated by hydration of the radical cations as reported in recent work (Y. Sun et al. , Phys. Chem. Chem. Phys. , 2018, 20 , 27510). As gas-phase hydration reactions closely mimic the actual scenario for guanine radical cations in double-stranded DNA, exploration of subsequent reactions within their water complexes can provide an insight into the resulting oxidative damage to nucleosides. Herein guided-ion beam mass spectrometry experiment and direct dynamics trajectory simulations were carried out to examine prototype complexes of the 9-methylguanine radical cation with one and two water ligands ( i.e. , 9MG˙ + ·(H 2 O) 1–2 ) in the gas phase, wherein the complexes were activated by collisional activation in the experiment and by thermal excitation at high temperatures in the simulations. Guided by mass spectroscopic measurements, trajectory results and reaction potential energy surface, three reaction pathways were identified. The first two reaction pathways start with H-atom abstraction from water by the O6 and N7 atoms in 9MG˙ + and are referred to as HA O6 and HA N7 , respectively. The primary products of HA O6 and HA N7 reactions, including [9MG + H O6 ] + /[9MG + H N7 ] + and ˙OH, react further to either form [8OH-9MG + H O6 ]˙ + and [8OH-9MG + H N7 ]˙ + via C8-hydroxylation or form radical cations of 6- enol -guanine (6- enol -G˙ + ) and 7H-guanine (7HG˙ + ) via S N 2-type methanol elimination. The third reaction pathway corresponds to the formation of 8OH-9MG + by H elimination from the complex, referred to as HE. Among these product channels, [8OH-9MG + H N7 ]˙ + has the most favorable formation probability, especially in the presence of additional water molecules. This product may serve as a preceding structure to the 8-oxo-7,8-dihydroguanine lesion in DNA and has implications for health effects of radiation exposure and radiation therapy. 

Humans are adept in simultaneously following multiple goals, but the neural mechanisms for maintaining specific goals and distinguishing them from other goals are incompletely understood. For short time scales, working memory studies suggest that multiple mental contents are maintained by theta-coupled reactivation, but evidence for similar mechanisms during complex behaviors such as goal-directed navigation is scarce. We examined intracranial electroencephalography recordings of epilepsy patients performing an object-location memory task in a virtual environment. We report that large-scale electrophysiological representations of objects that cue for specific goal locations are dynamically reactivated during goal-directed navigation. Reactivation of different cue representations occurred at stimulus-specific hippocampal theta phases. Locking to more distinct theta phases predicted better memory performance, identifying hippocampal theta phase coding as a mechanism for separating competing goals. Our findings suggest shared neural mechanisms between working memory and goal-directed navigation and provide new insights into the functions of the hippocampal theta rhythm.