Machine-learning models for activity class prediction: A comparative study of feature selection and classification algorithms

PurposeMachine-learning (ML) approaches have been repeatedly coupled with raw accelerometry to classify physical activity classes, but the features required to optimize their predictive performance are still unknown. Our aim was to identify appropriate combination of feature subsets and prediction algorithms for activity class prediction from hip-based raw acceleration data.MethodsThe hip-based raw acceleration data collected from 27 participants was split into training (70 %) and validation (30 %) subsets. A total of 206 time- (TD) and frequencydomain (FD) features were extracted from 6-second non-overlapping windows of the signal. Feature selection was done using seven filter-based, two wrapper-based, and one embedded algorithm, and classification was performed with artificial neural network (ANN), support vector machine (SVM), and random forest (RF). For every combination between the feature selection method and the classifiers, the most appropriate feature subsets were found and used for model training within the training set. These models were then validated with the left-out validation set.ResultsThe appropriate number of features for the ANN, SVM, and RF ranged from 20 to 45. Overall, the accuracy of all the three classifiers was higher when trained with feature subsets generated using filter-based methods compared with when they were trained with wrapper-based methods (range: 78.1 %–88 % vs. 66 %–83.5 %). TD features that reflect how signals vary around the mean, how they differ with one another, and how much and how often they change were more frequently selected via the feature selection methods.ConclusionsA subset of TD features from raw accelerometry could be sufficient for ML-based activity classification if properly selected from different axes.     

应用 σ度量对深圳市龙华区三家综合医院临床生化检验结果互认的评价研究

目的　探讨 σ度量在深圳市龙华区三家综合医院临床生化检验结果互认中的应用价值。方法　收集三家医院 2019年广东省临检中心常规化学，特殊蛋白前两次的室间质评（ EQA）样本，经混匀并分割成比对样本送至各评价实验室进行检测；以卫生部临检中心（ NCCL)标准和生物学变异导出的允许总误差（ TEa%）为质量规范，通过加权法计算出各项目 2019年 1～ 6月期间室内质控在控数据的不精密度，以 2018年参加卫生部室间质评（ EQA）数据作为偏倚（ bias%）来源，计算三家医院 29个检验项目的 σ度量值并进行性能评价，利用质量目标指数（ QGI)提出改善方法；将评价项目的 σ度量值与直接比对试验进行比较，通过统计分析确定 σ截断值（ cut-off）。结果　①三家医院 29个生化项目的性能并不一致，将项目中、高值的 σ值进行配对 t检验，其 t值分别为 2.28，3.01和 0.74，按 α =0.05水准判断得出医院 A与 B，医院 A与 C间的性能差异有统计学意义，医院 B与 C间的性能差异无统计学意义，医院 A项目性能要优于 B，C医院。评估的 87个项目中有 53项 σ平均值 <6，根据 QGI判断，3个项目需优先改善准确度，10个项目需同步改善准确度和精密度，另 40个项目需优先改善精密度。②对医院间项目比对通过率进行配对 χ2检验，其 χ2分别为 5.33，6.25和 2.5，按α=0.05水准判断， B与 A医院比对通过率低于 C与 A医院、 C与 B医院，差异有统计学意义； C与 A医院比对通过率跟 C与 B医院差异无统计学意义。③将 σ度量值与直接比对试验进行比较，经配对 χ2检验，当 σ cut-off值取≥ 2σ， ≥ 3σ，≥ 4σ，≥ 5σ和≥ 6σ时，其 χ2分别为 7.20，0.00，9.09，15.06和 16.06，按 χ20.05，1=3.84，α=0.05水准判断，当σcut-off值取≥ 3σ时，两种比对方法的差异无统计学意义。结论　σ度量值可以单独用于评价医院间生化检验结果的互认，在判断项目是否具有可比性时，可考虑将 σ cut-off值设为≥ 3σ。     

Electrophysiological evidence of altered facial expressions recognition in Alzheimer’s disease: A comprehensive ERP study

ObjectivesThe present study aims to evaluate the amplitude and latency of event-related potentials (ERPs) P100, N170, VPP and N230 in individuals with Alzheimer's disease (AD) compared to healthy elderly controls, using a passive viewing task of emotional facial expressions.MethodsTwenty-four individuals with mild to moderate AD and 23 demographically matched healthy elderly controls were included in the study. ERP P100, N170, VPP and N230 amplitude and latency values were compared between groups.ResultsThe categorization of emotional facial expressions was intact; yet, increased P100 amplitude and latency, decreased N170 amplitude, and increased VPP amplitude were observed in AD compared to controls. Increased N230 amplitude and latency were observed in response to angry expressions, while neutral expressions elicited decreased amplitude and latency.ConclusionsIncreased P100 amplitude and latency may reflect reduced amygdala volume and disruptions in the visual system, while decreased N170 and increased VPP amplitudes may reflect impaired perceptual processing, mitigated by a greater involvement of prefrontal areas for task performance in AD.SignificanceThis study is the first to report a complex pattern of ERPs to emotional facial expressions in individuals with AD.     

Neurofeedback and neuroplasticity of visual self-processing in depressed and healthy adolescents: A preliminary study

Adolescence is a neuroplastic period for self-processing and emotion regulation transformations, that if derailed, are linked to persistent depression. Neural mechanisms of adolescent self-processing and emotion regulation ought to be targeted via new treatments, given moderate effectiveness of current interventions. Thus, we implemented a novel neurofeedback protocol in adolescents to test the engagement of circuits sub-serving self-processing and emotion regulation.MethodsDepressed (n = 34) and healthy (n = 19) adolescents underwent neurofeedback training using a novel task. They saw their happy face as a cue to recall positive memories and increased displayed amygdala and hippocampus activity. The control condition was counting-backwards while viewing another happy face. A self vs. other face recognition task was administered before and after neurofeedback training.ResultsAdolescents showed higher frontotemporal activity during neurofeedback and higher amygdala and hippocampus and hippocampi activity in time series and region of interest analyses respectively. Before neurofeedback there was higher saliency network engagement for self-face recognition, but that network engagement was lower after neurofeedback. Depressed youth exhibited higher fusiform, inferior parietal lobule and cuneus activity during neurofeedback, but controls appeared to increase amygdala and hippocampus activity faster compared to depressed adolescents.ConclusionsNeurofeedback recruited frontotemporal cortices that support social cognition and emotion regulation. Amygdala and hippocampus engagement via neurofeedback appears to change limbic-frontotemporal networks during self-face recognition. A placebo group or condition and contrasting amygdala and hippocampus, hippocampi or right amygdala versus frontal loci of neurofeedback, e.g. dorsal anterior cingulate cortex, with longer duration of neurofeedback training will elucidate dosage and loci of neurofeedback in adolescents.     

White matter correlates of different aspects of facial affect recognition impairment following traumatic brain injury

Although facial affect recognition deficits are well documented in individuals with moderate-to-severe traumatic brain injury (TBI), little research has examined the neural mechanisms underlying these impairments. Here, we use diffusion tensor imaging (DTI), specifically the scalars fractional anisotropy (FA), mean diffusivity (MD), and radial diffusivity (RD), to examine relationships between regional white-matter integrity and two facial affect sub-skills: perceptual affect recognition abilities (measured by an affect matching task) and verbal categorization of facial affect (measured by an affect labeling task). Our results showed that, within the TBI group, higher levels of white-matter integrity in tracts involved in affect recognition (inferior fronto-occipital, inferior longitudinal, and uncinate fasciculi) were associated with better performance on both tasks. Verbal categorization skills were specifically and positively correlated with integrity of the left uncinate fasciculus. Moreover, we observed a striking lateralization effect, with perceptual abilities having an almost exclusive relationship with integrity of right hemisphere tracts, while verbal abilities were associated with both left and right hemisphere integrity. The findings advance our understanding of the neurobiological mechanisms that underlie subcomponents of facial affect recognition and lead to different patterns of facial affect recognition impairment in adults with TBI.     

Emotion recognition from body movement and gesture in children with Autism Spectrum Disorder is improved by situational cues

BackgroundResearch shows people with Autism Spectrum Disorder (ASD) have poorer emotion recognition (ER) compared to their typically developing (TD) peers. However, it is not known whether this is the case when stimuli are limited to gesture and posture, and lack facial expressions.MethodFifty-four children with (n = 27) and without (n = 27) ASD, matched on age and gender, completed an ER task, that used dynamic stimuli. Processing style bias, Autistic-like-traits and empathy were also measured. With ER as the outcome variable, a multilevel logistic model was created.ResultsChildren with ASD were found to be significantly less accurate in identifying emotions, compared to the control group. Presence of situational cues aided both groups. Autistic-like-traits and empathy were found to correlate too highly with the diagnosed condition to use in the multilevel model. Processing style did not significantly impact ER ability.ConclusionsThis study supports previous research which finds ER ability in people with ASD to be poorer than that of TD peers and that situational cues can aid ER ability. Importantly, the latter is true for people with ASD. The implication of these findings are programmes that aim to improve ER should consider using cues. Limitations of the study are discussed.     

基于化学模式识别技术提升参威骨痹片的质量标准并监测其生产中质控指标的转移率变化

目的:提升参威骨痹片的质量标准,初步探索其质量控制指标成分在批间含量差异较大的原因。方法:采用HPLC建立参威骨痹片的指纹图谱,以Diamonsil C18(4. 6 mm×250 mm,5μm)为色谱柱,流动相乙腈(A)-0. 1%磷酸水溶液(B)梯度洗脱(0～5 min,10%A;5～15 min,10%～12%A;15～30 min,12%～26%A;30～43 min,26%～31%A,43～50 min,31%～40%A,50～70 min,40%～55%A;70～84 min,55%～72. 5%A),检测波长230 nm。以共有峰为自变量绘制正交偏最小二乘法-判别分析-变量重要性投影(OPLS-DA-VIP)图,将共有峰对该制剂各批次间指纹图谱差异的贡献度量化,寻找差异较大的色谱峰,结合相关文献,筛选出与参威骨痹片临床适应症相关的成分并进行其含量测定的专属性试验,最终选定质控指标。通过HPLC-二极管阵列检测器(DAD)同时对本品及其生产过程中间体中质控指标进行测定,检测波长236,276,230,322 nm,其他条件同HPLC指纹图谱检测方法。结果:HPLC指纹图谱共标定了26个共有峰,各批次样品指纹图谱与对照指纹图谱的相似度均≥0. 950。优选出马钱苷酸、龙胆苦苷、芍药苷、蛇床子素为参威骨痹片的质控指标,四者的平均质量分数分别为161. 02,401. 80,255. 54,80. 68μg·g-1。结论:所建立的指纹图谱及多指标定量分析方法稳定、可靠,可用于参威骨痹片的质量控制。原料药批间质控指标成分含量差异和生产过程中间体的质控方法不够完善是引起该制剂批间质控指标成分含量差异较大的主要原因。     

An electronic travel guide for visually impaired – vehicle board recognition system through computer vision techniques

Abstract

Purpose: In the context of assistive technology, mobility takes the meaning of “moving safely, gracefully, and comfortably”.The aim of this article is to provide a system which will be a convenient means of navigation for the Visually Impaired people, in the public transport system.

Method: A blind regular commuter who travels by public transport facility finds difficulty in identifying the vehicle that is nearing the stop. Hence, a real-time system that dynamically identifies the nearing vehicle and informs the commuters is necessary. This paper proposes such a system namely the “Vehicle Board Recognition System” (VBRS). Computer Vision techniques such as segmentation, object recognition, text detection and optical character recognition are utilized to build the system, which will detect, analyze, derive and communicate the information to the passengers.

Results: Thanks to the rapid development in technology, there are several navigation systems both hand held and wearable, available to help visually impaired (VI) people move comfortably both indoor and outdoor. Many blind people are not comfortable in using these devices or they are not affordable for them. Thus the proposed system gives them the comfort of navigation.

Conclusion: This system can be installed in the bus stop to assist the Visually Impaired, from externally rather than their hand held or wearable assistive devices.

• Implications for rehabilitation

• This proposed system will help the visually impaired to

• ensure secure navigation

• be independent of the others

• develop self confidence.

• overcome the training, affordability of wearable/ handheld devices.