Search

The three-dimensional swimming tracks of motile microorganisms can be used to identify their species, which holds promise for the rapid identification of bacterial pathogens. The tracks also provide detailed information on the cells' responses to external stimuli such as chemical gradients and physical objects. Digital holographic microscopy (DHM) is a well-established, but computationally intensive method for obtaining three-dimensional cell tracks from video microscopy data. We demonstrate that a common neural network (NN) accelerates the analysis of holographic data by an order of magnitude, enabling its use on single-board computers and in real time. We establish a heuristic relationship between the distance of a cell from the focal plane and the size of the bounding box assigned to it by the NN, allowing us to rapidly localise cells in three dimensions as they swim. This technique opens the possibility of providing real-time feedback in experiments, for example by monitoring and adapting the supply of nutrients to a microbial bioreactor in response to changes in the swimming phenotype of microbes, or for rapid identification of bacterial pathogens in drinking water or clinical samples.

IntroductionA proposed Diagnostic AI System for Robot-Assisted Triage (“DAISY”) is under development to support Emergency Department (“ED”) triage following increasing reports of overcrowding and shortage of staff in ED care experienced within National Health Service, England (“NHS”) but also globally. DAISY aims to reduce ED patient wait times and medical practitioner overload. The objective of this study was to explore NHS health practitioners' perspectives and attitudes towards the future use of AI-supported technologies in ED triage.MethodsBetween July and August 2022 a qualitative-exploratory research study was conducted to collect and capture the perceptions and attitudes of nine NHS healthcare practitioners to better understand the challenges and benefits of a DAISY deployment. The study was based on a thematic analysis of semi-structured interviews. The study involved qualitative data analysis of the interviewees' responses. Audio-recordings were transcribed verbatim, and notes included into data documents. The transcripts were coded line-by-line, and data were organised into themes and sub-themes. Both inductive and deductive approaches to thematic analysis were used to analyse such data.ResultsBased on a qualitative analysis of coded interviews with the practitioners, responses were categorised into broad main thematic-types, namely: trust; current practice; social, legal, ethical, and cultural concerns; and empathetic practice. Sub-themes were identified for each main theme. Further quantitative analyses explored the vocabulary and sentiments of the participants when talking generally about NHS ED practices compared to discussing DAISY. Limitations include a small sample size and the requirement that research participants imagine a prototype AI-supported system still under development. The expectation is that such a system would work alongside the practitioner. Findings can be generalisable to other healthcare AI-supported systems and to other domains.DiscussionThis study highlights the benefits and challenges for an AI-supported triage healthcare solution. The study shows that most NHS ED practitioners interviewed were positive about such adoption. Benefits cited were a reduction in patient wait times in the ED, assistance in the streamlining of the triage process, support in calling for appropriate diagnostics and for further patient examination, and identification of those very unwell and requiring more immediate and urgent attention. Words used to describe the system were that DAISY is a “good idea”, “help”, helpful, “easier”, “value”, and “accurate”. Our study demonstrates that trust in the system is a significant driver of use and a potential barrier to adoption. Participants emphasised social, legal, ethical, and cultural considerations and barriers to DAISY adoption and the importance of empathy and non-verbal cues in patient interactions. Findings demonstrate how DAISY might support and augment human medical performance in ED care, and provide an understanding of attitudinal barriers and considerations for the development and implementation of future triage AI-supported systems.

Competitive video game playing, an activity called esports, is increasingly popular to the point that there are now many professional competitions held for a variety of games. These competitions are broadcast in a professional manner similar to traditional sports broadcasting. Esports games are generally fast paced, and due to the virtual nature of these games, camera positioning can be limited. Therefore, knowing ahead of time where to position cameras, and what to focus a broadcast and associated commentary on, is a key challenge in esports reporting. This gives rise to moment-to-moment prediction within esports matches which can empower broadcasters to better observe and process esports matches. In this work we focus on this moment-to-moment prediction and in particular present techniques for predicting if a player will die within a set number of seconds for the esports title Dota 2. A player death is one of the most consequential events in Dota 2. We train our model on ‘telemetry’ data gathered directly from the game itself, and position this work as a novel extension of our previous work on the challenge. We use an enhanced dataset covering 9,822 Dota 2 matches. Since the publication of our previous work, new dataset parsing techniques developed by the WEAVR project enable the model to track more features, namely player status effects, and more importantly, to operate in real time. Additionally, we explore two new enhancements to the original model: one data-based extension and one architectural. Firstly we employ learnt embeddings for categorical features, e.g. which in game character a player has selected, and secondly we explicitly model the temporal element of our telemetry data using recurrent neural networks. We find that these extensions and additional features all aid the predictive power of the model achieving an F1 score of 0.54 compared to 0.17 for our previous model (on the new data). We improve this further by experimenting with the length of the time-series in the input data and find using 15 time steps further improves the F1 score to 0.62. This compares to F1 of 0.1 for a standard RNN on the same task. Additionally a deeper analysis of the Time to Die model is carried out to assess its suitability as a broadcast aid.

Bullying victimization is a pervasive problem nationwide and is related to students' psychological distress, including increased loneliness, anxiety, depression, helplessness, and suicidal behaviors. Importantly, not all students respond to peer victimization in the same way. This study examined the effectiveness of students' coping strategies in response to bullying as a function of their grade level, gender, and form of victimization. Students in grades five through twelve (N = 2627) completed surveys assessing coping strategy effectiveness and associated emotional distress related to self-reported peer victimization. In general, findings indicated that the most effective coping strategies in response to bullying victimization were support-seeking strategies, use of humor, and cognitive restructuring. However, effectiveness for all strategies (with the exception of peer support) varied by grade level, gender, and form of victimization. Strategies were also more effective for relational victimization than they were for physical victimization. Moreover, although students identified several effective coping strategies, only the use of cognitive restructuring was related to lower levels of associated emotional distress, an important component of effective coping. As such, bullying prevention programs should target students' psychological processes in developing effective interventions. Implications for prevention and intervention efforts are discussed.

Ram semen traits play a significant role in conception outcomes, which in turn may influence reproductive efficiency and the overall productivity and profitability of sheep enterprises. Since hundreds of ewes may be inseminated from a single ejaculate, it is important to evaluate semen quality prior to use in sheep breeding programs. Given that semen traits have been found to be heritable, genetic variation likely contributes to the variability observed in these traits. Identifying such genetic variants could provide novel insights into the molecular mechanisms underlying variability in semen traits. Therefore, this study aimed to identify quantitative trait loci (QTLs) associated with semen traits in Merino sheep. A genome-wide association study (GWAS) was undertaken using 4506 semen collection records from 246 Merino rams collected between January 2002 and May 2021. The R package RepeatABEL was used to perform a GWAS for semen volume, gross motility, concentration, and percent post-thaw motility. A total of 35 QTLs, located on 16 Ovis aries autosomes (OARs), were significantly associated with either of the four semen traits in this study. A total of 89, 95, 33, and 73 candidate genes were identified, via modified Bonferroni, within the QTLs significantly associated with volume, gross motility, concentration, and percent post-thaw motility, respectively. Among the candidate genes identified, SORD, SH2B1, and NT5E have been previously described to significantly influence spermatogenesis, spermatozoal motility, and high percent post-thaw motility, respectively. Several candidate genes identified could potentially influence ram semen traits based on existing evidence in the literature. As such, validation of these putative candidates may offer the potential to develop future strategies to improve sheep reproductive efficiency. Furthermore, Merino ram semen traits are lowly heritable (0.071–0.139), and thus may be improved by selective breeding.

Background: Perinatal stroke is the most common cause of unilateral cerebral palsy. Mechanisms of post-stroke developmental plasticity in children are poorly understood. To better understand the relationship between functional connectivity and disability, we used resting-state fMRI to compare sensorimotor connectivity with clinical dysfunction. Methods: School-aged children with periventricular venous infarction (PVI) and unilateral cerebral palsy were compared to controls. Resting-state BOLD signal was acquired on 3 T MRI and analyzed using CONN in SPM12. Functional connectivity was computed between S1, M1, supplementary motor area (SMA), and thalamus of the left/non-lesioned and right/lesioned hemisphere. Primary outcome was connectivity expressed as a Fisher-transformed correlation coefficient. Motor function was measured using the Assisting Hand Assessment (AHA), and Melbourne Assessment (MA). Proprioceptive function was measured using a robotic position matching task (VarXY). Results: Participants included 15 PVI and 21 controls. AHA and MA in stroke patients were negatively correlated with connectivity (increased connectivity = poorer performance). Position sense was inversely correlated with connectivity (increased connectivity = improved performance) between the non-lesioned S1 and thalamus/SMA. In controls, VarXY was positively correlated with connectivity between the thalamus and bilateral sensorimotor regions. Conclusions: Resting state fMRI measures of sensorimotor connectivity are associated with clinical sensorimotor function in children with unilateral cerebral palsy secondary to PVI. Greater insight into understanding reorganization of brain networks following perinatal stroke may facilitate personalized rehabilitation. Keywords: Functional MRI, Sensorimotor resting-state networks, Periventricular venous infarction, Perinatal stroke, Network connectivity, Proprioceptive function, Cerebral palsy

Bi-factor confirmatory factor models have been influential in research on cognitive abilities because they often better fit the data than correlated factors and higher-order models. They also instantiate a perspective that differs from that offered by other models. Motivated by previous work that hypothesized an inherent statistical bias of fit indices favoring the bi-factor model, we compared the fit of correlated factors, higher-order, and bi-factor models via Monte Carlo methods. When data were sampled from a true bi-factor structure, each of the approximate fit indices was more likely than not to identify the bi-factor solution as the best fitting. When samples were selected from a true multiple correlated factors structure, approximate fit indices were more likely overall to identify the correlated factors solution as the best fitting. In contrast, when samples were generated from a true higher-order structure, approximate fit indices tended to identify the bi-factor solution as best fitting. There was extensive overlap of fit values across the models regardless of true structure. Although one model may fit a given dataset best relative to the other models, each of the models tended to fit the data well in absolute terms. Given this variability, models must also be judged on substantive and conceptual grounds.