Search

Attention-deficit/hyperactivity disorder (ADHD) is a relatively commonly occurring neurodevelopmental disorder affecting approximately 5% of children and young people. The neurobiological mechanisms of ADHD are proposed to particularly center around increased dopamine receptor availability related to associated symptoms of reduced attention regulation and impulsivity. ADHD is also persistent across the lifespan and associated with a raft of impulsive and health-risk behaviors including substance abuse and smoking. Research highlighting the potentially significant levels of monoamine oxidase (MAO) inhibitory properties in tobacco smoke and e-cigarettes may provide a mechanism for increased tobacco smoke dependence among those with ADHD, in addition to the role of nicotine.AimThis scoping review aimed to establish evidence for the above neurobiological pathway between smoking and ADHD symptom-alleviation or “self-medication” with the inclusion of the mechanism of MAO-inhibitors indirect increasing dopamine in the brain.MethodologyScoping review methodologies were employed in this review selected to synthesize multiple sources of empirical research to identify current gaps in the knowledge base and identify key characteristics of research data related to a phenomenon. Databases searched included OVID MEDLINE(R), Embase, Cochrane, PsycINFO and SCOPUS limited to 2000 onward and empirically validated, peer-reviewed research.FindingsThere is support for the role of MAO-inhibition on greater reinforcement of smoking for individuals with ADHD through a greater impact on dopaminergic availability than nicotine; potentially moderating ADHD symptoms.ConclusionGreater support for a “self-medication” model of ADHD and smoking includes not only nicotine but also MAO-inhibitors as dopamine agonists contained in cigarettes and e-cigarettes.

In this article we discuss how to use orgmode for writing and weaving data analysis.

Energy-efficient machine-learning and physical unclonable function (PUF) has drawn significant attention for Internet-of-Things (IoT) application in wake-up detection for bandwidth/computation reduction and privacy protection at sensor node or autonomous device. A machine-learning and PUF engine for IoT applications is presented in this paper with a current mirror cross-bar (CMCB) being a shared core circuit for both functions, leading to reduction in overhead area by $48.5\times $ . A novel dimension expansion technique is proposed to increase weight matrix dimension beyond the physically implemented array with small hardware and energy overhead. A signed multiply–accumulation is realized in CMCB with differential current path and 2-phase conversion. The proposed engine achieves an error rate of 6.34% on MNIST digit recognition task with an energy efficiency of 2.86 TOPS/W. The PUF achieves a native bit error rate of 2.3% across corners and extremely low area per challenge response pair (CRP) of $4.17\times 10^{-59}\,\,\mu \text{m}^{2}$ /CRP due to exponentially more CRP enabled by ternary input mode.

Objectives: This paper provides a discussion about the potential scope of applicability of Artificial Intelligence methods within the telehealth domain. These methods are focussed on clinical needs and provide some insight to current directions, based on reports of recent advances. Methods: Examples of telehealth innovations involving Artificial Intelligence to support or supplement remote health care delivery were identified from recent literature by the authors, on the basis of expert knowledge. Observations from the examples were synthesized to yield an overview of contemporary directions for the perceived role of Artificial Intelligence in telehealth. Results: Two major focus areas for related contemporary directions were established. These were first, quality improvement for existing clinical practice and service delivery, and second, the development and support of new models of care. Case studies from each focus area have been chosen for illustration purposes. Conclusion: Examples of the role of Artificial Intelligence in delivery of health care remotely include use of tele-assessment, tele-diagnosis, tele-interactions, and tele-monitoring. Further developments of underlying algorithms and validation of methods will be required for wider adoption. Certain key social and ethical considerations also need consideration more generally in the health system, as Artificial-Intelligence-enabled-telehealth becomes more commonplace.

Purpose The purpose of this paper is to assess policy and management of hypertension and diabetes in Ghana. Design/methodology/approach The authors conducted 26 key informant interviews with policy makers, payers, providers and participants from advocacy groups associated with the management of hypertension and diabetes both at national and district levels in Ghana; conducted focus group discussions with a total of 18 hypertensive and diabetic patients; and analyzed documentation detailing activities that have been undertaken in response to the management of hypertension and diabetes in Ghana. The authors then conducted a content analysis after combining the three sources of information. Findings Using a stepwise policy process, the Ghanaian health sector has developed overarching policies and strategies on management of diabetes, hypertension, other non-communicable diseases, tobacco, alcohol and nutrition-related issues. Availability of funds and over-concentration on communicable diseases are the main barriers to the implementation of policies. Besides, response of the health sector to address the problems of hypertension and diabetes is focused more on clinical management than prevention; while the structures and processes to manage diabetes and hypertension is available at all levels of service delivery, more needs to be done on preventive aspects. Originality/value This is the first study in Ghana to combine individual interviews, focus groups and document analysis to review policy development, implementation and response activities on the management of hypertension and diabetes. The authors believe that the evidence from this research will inform future policy initiatives on hypertension and diabetes management in Ghana.

The last decade (2010-2019) has seen calls to action to improve the prescribing practice of junior doctors. An in-depth investigation into the causes of prescribing errors by foundation trainees in relation to their medical education (the EQUIP study) in the UK reported a prescription error rate of 8.9% for all prescribed medicines, and although that is a UK study, there are similarities with New Zealand prevocational training programmes. The EQUIP study revealed that existing teaching strategies are not working. To believe a single intervention will prevent most prescribing errors is simplistic, and for improvement to occur, new prescribers need to learn from their mistakes. Traditionally, the education of junior doctors has focused on their competence and professional registration requirements. Working in healthcare is collective and multidisciplinary, and errors occur through human and system factors.

Cronbach's alpha and omega are units of reliability of a set of items in a questionnaire. We discuss here how to calculate Cronbach's alpha and how to interpret the values.

Evidence gap maps are visual matrices to show gaps in available evidence about a research questions. In order to construct an evidence gap map, you should select an answerable research question, search available literature, select studies, assess their quality or critically appraise their quality, and then tally each type of study for each intervention-outcome or exposure-outcome pair. After all the cells are filled or left blank, the blank cells in such a framework show the gaps in the evidence. We describe here the processes and do a demonstration step by step

The goal of this paper is to present a tutorial on structural equation modelling (“SEM”). SEM is a combination of multivariate linear regression and path analysis models. We will discuss path analysis, measurement models, measurement invariance and when or how to use them, twin studies, and longitudinal data analysis. In this tutorial, we shall use the free and open source package “lavaan” in R.

This is a conceptual introduction to ggplot2 in Rstudio. We take an available data set in R and work step by step to identify patterns and produce publishable quality graphics using the principles of grammar of graphics.

In this paper, we introduce the concepts of critically reading research papers and writing of research proposals and reports. Research methods is a general term that includes the processes of observation of the world around the researcher, linking background knowledge with foreground questions, drafting a plan of collection of data and framing theories and hypotheses, testing the hypotheses, and finally, drafting or writing the research to evoke new knowledge. These processes vary with the themes and disciplines that the researcher engages in; nevertheless, common motifs can be found. In this paper, we propose three methods are interlinked: a deductive reasoning process where the structure of the thought can be captured critically; an inductive reasoning method where the researcher can appraise and generate generalisable ideas from observations of the world; and finally, abductive reasoning method where the world can be explained or the phenomena observed can be explained or be accounted for. This step or reasoning is also about framing theories, testing and challenging established knowledge or finding best theories and how theories best fit the observations. We start with a discussion of the different types of statements that one can come across in any scholarly literature or even in lay or semi-serious literature, appraise them, and identify arguments from non-arguments, and explanations from non-explanations. Then we outline three strategies to appraise and identify reasonings in explanations and arguments. We end with a discussion on how to draft a research proposal and a reading/archiving strategy of research.

Commercial machine learning accelerators like Intel neural Compute Stick 2 (NCS2) enable efficient inference on otherwise low resource edge devices. However, these accelerators are also ex- posed to new threats leveraging physical access. In this paper, we present the first results demonstrating practical electromagnetic side-channel attack on NCS2, allowing secret weight recovery from executed models National Research Foundation (NRF) This research is supported by the National Research Foundation, Singapore, under its National Cybersecurity Research & Development Programme / Cyber-Hardware Forensic & Assurance Evaluation R&D Programme (Award: NRF2018NCR-NCR009-0001)

With the advent of modern engineered fibers and highly developed electronics a new variety of textile material has been evolved which is known as smart textiles. A smart textile generally has five major functions, that is, sensors, actuators, data processing, energy supply, and communication. Development of thin, flexible, and highly sensitive sensors has made it possible to detect the problems faced by the human during sickness or while participating in sports. This article reviews the development in all five areas and their future directions. Technologies such as printing, coating, lamination, and embroidery have been used for producing thin, flexible sensors on textile substrate, or plastic which helped in producing wearable smart textiles. The textile yarns, fabrics have been converted into conductive materials by chemical methods, mixing metallic fibers and other methods. Human biomonitoring is one of the most important uses of smart textiles, which either reacts or communicates messages depending on the values sensed by the sensors.

Edge deep learning accelerators are optimised hard ware to enable efficient inference on the edge. The models deployed on these accelerators are often proprietary and thus sensitive for commercial and privacy reasons. In this paper, we demonstrate practical vulnerability of deployed deep learning models to timing side-channel attacks. By measuring the execution time of the inference, the adversary can determine and reconstruct the model from a known family of well known deep learning model and then use available techniques to recover remaining hyperparameters. The vulnerability is validated on Intel Compute Stick 2 for VGG and ResNet family of models. Moreover, the presented attack is quite devastating as it can be performed in a cross-device setting, where adversary profiles constructed on a legally own device can be used to exploit the victim device with a single query and still can achieve near perfect success rate. National Research Foundation (NRF) Accepted version This research is supported in parts by the National Research Foundation, Singapore, under its National Cybersecurity Research & Development Programme / Cyber-Hardware Forensic & Assurance Evaluation R&D Programme (Award: NRF2018NCR-NCR009-0001).

Intra-cortical Brain Machine Interfaces (iBMIs) with wireless capability could scale the number of recording channels by integrating an intention decoder to reduce data rates. However, the need for frequent retraining due to neural signal non-stationarity is a big impediment. This paper presents an alternate paradigm of online reinforcement learning (RL) with a binary evaluative feedback in iBMIs to tackle this issue. This paradigm eliminates time-consuming calibration procedures. Instead, it relies on updating the model on a sequential sample-by-sample basis based on an instantaneous evaluative binary feedback signal. However, batch updates of weight in popular deep networks is very resource consuming and incompatible with constraints of an implant. In this work, using offline open-loop analysis on pre-recorded data, we show application of a simple RL algorithm - Banditron -in discrete-state iBMIs and compare it against previously reported state of the art RL algorithms – Hebbian RL, Attention gated RL, deep Q-learning. Owing to its simplistic single-layer architecture, Banditron is found to yield at least two orders of magnitude of reduction in power dissipation compared to state of the art RL algorithms. At the same time, post-hoc analysis performed on four pre-recorded experimental datasets procured from the motor cortex of two non-human primates performing joystick-based movement-related tasks indicate Banditron performing significantly better than state of the art RL algorithms by at least 5%, 10%, 7% and 7% in experiments 1, 2, 3 and 4 respectively. Furthermore, we propose a non-linear variant of Banditron, Banditron-RP, which gives an average improvement of 6%, 2% in decoding accuracy in experiments 2,4 respectively with only a moderate increase in power consumption.

In this review of how novices can develop mastery, I will summarise the key research from educational studies to the pathway of mastery. According to the five stage theory of development of mastery by Stuart and Hubert Dreyfus, everyone begin as novices. Novices in their journey to develop expertise require heuristics and rules that are context free to develop skills. With more experience and with deliberate practice with graduated challenges and feedbacks to correct errors, the novice then reaches a stage where the novice then contextualises the heuristics and rules. This is the stage where the novice then starts developing “tacit” knowledge that will enable the novice to start contextualising the problems that the novice can start to solve. In the third stage of developing mastery, the novice becomes a competent practitioner where the novice starts building an emotional attachments as they gain more experiences with a range of situations and therefore develop repertoire of skills and positions. However, at this stage, since the competent practitioner has now developed a vast range of skills and knowledge about problem solving, understanding which technique or which strategy to adopt in what situations become challenging. This stage is emotionally exhausting for the competent practitioner but the competent practitioner is now emotionally connected to the knowledge process or solving problems. Thus, if a problem is solved, the competent practitoner is elated and if the problem is not solved or the competent practitioner fails, they are emotionally upset. With more practice and development and gain of skills, experiences, the competent practitioner enters the stage of expertise. In this stage, they still analyse and use heuristics and calculations to solve problems, with even stronger emotional attachment. Finally, in stage five, when the expert becomes a master, a sense of fluidity develops, where the master is aware of, but does not need to explicitly calculate or use heuristics to solve problems, instead a sense of creativity and flow develops in the master. Patricia Benner corroborated these concepts in the context of nursing and medical students and training of doctors. Here, the nurses and health providers undergo these five stages as they continue to develop mastery. Students start as novice nurses where they use rote learning and heuristics to solve problems. Then, as they gain more experiences in real wards and patient care, they learn how to contextualise the learned concepts in real world patient care. Over time, they develop rich repertoire of experiences and become competent practitioners to solve most day to day problems. Finally in the stage of mastery and expertise, they do not need to use the heuristics and rules anymore. These five-stage development of mastery can be mapped on to the concepts of cognitive development of mastery in learning that Benjamin Bloom proposed in 1956 working with school students and then he further developed as taxonomy of learning using a six-stage model. This is referred to as Bloom’s taxonomy of learning mastery where the most basic stage is “remember” where the learning objective is recalling information. This is really a low level but basic activity where the students gain knowledge by rote learning but not necessarily be able to contextualise it to their learned circumstances. In the second stage of development of mastery, once the students have demonstrated the mastery of rote learning, they show that they “understand” the core concepts. In the third stage, they are able to apply the learned concepts to new and unseen or situations that they had not previously experienced. In the fourth stage, they are able to analyse novel situations and can solve problems. In the final stage, they can create new knowledge. These stages of learning mastery that Bloom proposed map on to the five-stage model of mastery development that Dreyfus proposed and Benner validated. The question then, is, what is the mechanism by which this mastery is gained? What can a teacher do to develop expertise and mastery in a student? Bloom’s concept was that it should be based on initially providing lessons, then testing whether the student has learned the concepts, and if there are broken mental models that need to be addressed by correcting errors and providing new challenges. Each time the teacher presents a new, and slightly more difficult challenge to the student; the student attempts to solve it and gains valuable experience doing so. As each stage is reached, the teacher then progresses the student to the next stage of performance. In this way, the teacher traverses the five stage of mastery by creating a mastery rubric. We put these concepts in the context of how we can teach public health and evidence appraisal or critical thinking for students using this mastery rubric and graduated challenges. At the lowest stage, students are provided with lessons that involve rote learning and recall based assignments and formative assignments. In subsequent stages, we provide the students with more graduated challenges that they attempt to solve as they move up the ladder of expertise. Finally, the students are asked to develop a research proposal or comprehensively critique a body of knowledge.

GRADE provides a structure for clinicians and healthcare analysts to understand the quality of evidence for an outcome. An outcome can be associated with a number of studies, and quality appraisal in terms of whether additional evidence must be needed before a definitive conclusion about an intervention-outcome can be arrived at is done on the basis of conducting GRADE tables. GRADEpro as a software enables construction of grade tables, and has two parts. In part one of this series, we showed how to construct certainty of evidence tables. In this part, we discuss how to create summary of findings tables with GRADEpro for dichotomous outcome variables.

This is a step by step tutorial on how to use GRADEpro.

In this document, we describe the process of environmental health risk assessment, and illustrate a case of an environmental health risk assessment of nickel in drinking water in an Australian setting. We also provide some guidance on how to assess an environmental health risk assessment.

Applications in the Internet of Video Things (IoVT) domain have very tight constraints with respect to power and area. While neuromorphic vision sensors (NVS) may offer advantages over traditional imagers in this domain, the existing NVS systems either do not meet the power constraints or have not demonstrated end-to-end system performance. To address this, we improve on a recently proposed hybrid event-frame approach by using morphological image processing algorithms for region proposal and address the low-power requirement for object detection and classification by exploring various convolutional neural network (CNN) architectures. Specifically, we compare the results obtained from our object detection framework against the state-of-the-art low-power NVS surveillance system and show an improved accuracy of 82.16% from 63.1%. Moreover, we show that using multiple bits does not improve accuracy, and thus, system designers can save power and area by using only single bit event polarity information. In addition, we explore the CNN architecture space for object classification and show useful insights to trade-off accuracy for lower power using lesser memory and arithmetic operations., 4 pages, 2 figures

Due to increased adoption of IoT devices, Physically Unclonable Function (PUF) circuits are essential as a lightweight security primitive. However existing PUFs suffer from imperfect reliability due to environmental variation. In this paper, we propose a PUF-server paradigm in which the server hosts a temperature model of the PUF, and the PUF informs the server of its response, and temperature through an integrated sensor. The model can predict responses for two different strong PUFs across −45°C to 90°C and thus avoids losing challenge-response pairs(CRPs) owing to low mismatch. Simulations show the model-based reliability improves to 97.9% − 99.5% compared with conventional reliability of 83.9% − 86.3%. Conversely, a loss of 25% − 40% of the CRPs is observed in order to improve the conventional reliability to match the model-based reliability. These results are also shown to be robust against quantization errors of a temperature sensor.

Sensory information processing in robot skins currently rely on a centralized approach where signal transduction (on the body) is separated from centralized computation and decision-making, requiring the transfer of large amounts of data from periphery to central processors, at the cost of wiring, latency, fault tolerance and robustness. We envision a decentralized approach where intelligence is embedded in the sensing nodes, using a unique neuromorphic methodology to extract relevant information in robotic skins. Here we specifically address pain perception and the association of nociception with tactile perception to trigger the escape reflex in a sensorized robotic arm. The proposed system comprises self-healable materials and memtransistors as enabling technologies for the implementation of neuromorphic nociceptors, spiking local associative learning and communication. Configuring memtransistors as gated-threshold and -memristive switches, the demonstrated system features in-memory edge computing with minimal hardware circuitry and wiring, and enhanced fault tolerance and robustness., Sensory information processing in robots relies on a centralized approach with issues of wiring, fault-tolerance and latency. Here, the authors report a decentralized neuromorphic approach with self-healable memristive elements enabling intelligent sensations in a prototypical robotic nervous system.

Directed acyclic graphs are graphs that contain one directional arrows which connect the nodes within the graph structure, and where flow of information can be shown to flow from "past" to "future" along the direction of the arrows. These graphs are acyclic in the sense that no paths turn back on to the parent node as they are directed from a causal variable to an effect variable. In this paper, we discuss these graphs with respect to causal inference in Epidemiology and discuss ways of drawing our assumptions prior to our conclusions. Specifically, graphs will help us to identify biases and also help us to characterise counterfactual theories of causation

The primary objective of this paper is to build classification models and strategies to identify breathing sound anomalies (wheeze, crackle) for automated diagnosis of respiratory and pulmonary diseases. In this work we propose a deep CNN-RNN model that classifies respiratory sounds based on Mel-spectrograms. We also implement a patient specific model tuning strategy that first screens respiratory patients and then builds patient specific classification models using limited patient data for reliable anomaly detection. Moreover, we devise a local log quantization strategy for model weights to reduce the memory footprint for deployment in memory constrained systems such as wearable devices. The proposed hybrid CNN-RNN model achieves a score of $\text{66.31}\%$ on four-class classification of breathing cycles for ICBHI’17 scientific challenge respiratory sound database. When the model is re-trained with patient specific data, it produces a score of $\text{71.81}\%$ for leave-one-out validation. The proposed weight quantization technique achieves ${\approx}4 \times$ reduction in total memory cost without loss of performance. The main contribution of the paper is as follows: Firstly, the proposed model is able to achieve state of the art score on the ICBHI’17 dataset. Secondly, deep learning models are shown to successfully learn domain specific knowledge when pre-trained with breathing data and produce significantly superior performance compared to generalized models. Finally, local log quantization of trained weights is shown to be able to reduce the memory requirement significantly. This type of patient-specific re-training strategy can be very useful in developing reliable long-term automated patient monitoring systems particularly in wearable healthcare solutions.

The creation of useful categories from data is an important cognitive ability, and from the extensive research on categorization, it is now known that the brain has distinct systems for category learning. In this paper, we present the first spiking neural network (SNN) model of human category learning. Here categories are combinations of features - such categories are observed in the prefrontal cortex (PFC). The system follows an architecture commonly used to model the cortex - features are arranged in a topological 2D grid with short range excitatory connectivity and long range inhibitory connectivity - however, here, this architecture is used differently from other models to model higher level cognition. Earlier we presented an artificial neural network (ANN) model of category learning; however, here, a simpler model was adequate, as desired functionality emerges from the SNN dynamics. We identified the objectives that had to be fulfilled for the model to achieve the desired functionality, and performed a design space exploration (DSE) to identify the parameter range in which each of the objectives was fulfilled, and the parameter range for which the system exhibits good performance. Finally, we compared triphasic STDP (a variant of spike time dependant plasticity (STDP)) with standard STDP and observed that triphasic STDP exhibited quicker convergence.

Shallow feed-forward networks are incapable of addressing complex tasks such as natural language processing that require learning of temporal signals. To address these requirements, we need deep neuromorphic architectures with recurrent connections such as deep recurrent neural networks. However, the training of such networks demand very high precision of weights, excellent conductance linearity and low write-noise- not satisfied by current memristive implementations. Inspired from optogenetics, here we report a neuromorphic computing platform comprised of photo-excitable neuristors capable of in-memory computations across 980 addressable states with a high signal-to-noise ratio of 77. The large linear dynamic range, low write noise and selective excitability allows high fidelity opto-electronic transfer of weights with a two-shot write scheme, while electrical in-memory inference provides energy efficiency. This method enables implementing a memristive deep recurrent neural network with twelve trainable layers with more than a million parameters to recognize spoken commands with >90% accuracy., Accomplishing complex cognitive tasks such as speech recognition calls for artificial intelligence hardware with high computing precision. John et al. propose deep recurrent neural networks based on optoelectronic transition metal dichalcogenide memristors with high weight precision for in-memory computing.

We present a novel readout circuit for a ferro-magnetic Hall cross-bar based random number generator. The random orientation of magnetic domains are result of anomalous Hall-effect. These ferromagnetic Hall cross-bar structures can be integrated with the read out circuit to form a plug and play random number generator. The system can resolve up to 15– $20~\mu \text{V}$ Hall-voltages from Hall probe. Application of current densities around $10^{12} {A/m}^{2}$ through the Ferromagnetic Hall cross-bar produces random Hall-voltage on the output terminals. To amplify the weak Hall-voltages (10– $100~\mu \text{V}$ ) in the presence of DC offsets, a modulation scheme is used to up-convert the signal and a band-pass amplifier is used to amplify the modulated signal. The band-pass amplifier circuit, motivated by neural recording amplifier is designed in 65nm CMOS and consumes $126~\mu \text{W}$ of power from a 1.2 V supply. Further, we present a successive approximation algorithm and its embedded implementation to set the desired threshold for digitizing the amplified Hall-voltage in presence of signal drift. Experimental results show that the resulting system can tolerate drifts in voltage up to $440~\mu \text{V}$ .

In this chapter, we provide a smorgasbord of emergent technologies on the watch. We begin with an introduction to telehomecare; continue our discussions on the topic of the use of robots for medical and healthcare applications, artificial intelligence and its usage in health and medical care, and virtual reality and its applications in healthcare and medicine; and conclude with blockchain technology. We discuss these technologies with a view to their present and future potential.