Centroid training to achieve effective text classification

Traditional text classification technology based on machine learning and data mining techniques has made a big progress. However, it is still a big problem on how to draw an exact decision boundary between relevant and irrelevant objects in binary classification due to much uncertainty produced in the process of the traditional algorithms. The proposed model CTTC (Centroid Training for Text Classification) aims to build an uncertainty boundary to absorb as many indeterminate objects as possible so as to elevate the certainty of the relevant and irrelevant groups through the centroid clustering and training process. The clustering starts from the two training subsets labelled as relevant or irrelevant respectively to create two principal centroid vectors by which all the training samples are further separated into three groups: POS, NEG and BND, with all the indeterminate objects absorbed into the uncertain decision boundary BND. Two pairs of centroid vectors are proposed to be trained and optimized through the subsequent iterative multi-learning process, all of which are proposed to collaboratively help predict the polarities of the incoming objects thereafter. For the assessment of the proposed model, F1 and Accuracy have been chosen as the key evaluation measures. We stress the F1 measure because it can display the overall performance improvement of the final classifier better than Accuracy. A large number of experiments have been completed using the proposed model on the Reuters Corpus Volume 1 (RCV1) which is important standard dataset in the field. The experiment results show that the proposed model has significantly improved the binary text classification performance in both F1 and Accuracy compared with three other influential baseline models.

Perceived safety, quality and cultural competency of maternity care for culturally and linguistically diverse women in Queensland

Various policies, plans, and initiatives have been implemented to provide safe, quality, and culturally competent care to patients within Queensland’s healthcare system. A series of models of maternity care are available in Queensland that range from standard public care to private midwifery care. The current study aimed to determine whether identifying as Culturally or Linguistically Diverse (CALD) was associated with the perceived safety, quality, and cultural competency of maternity care from a consumer perspective, and to identify specific needs and preferences of CALD maternity care consumers. Secondary analysis of data collected in the Having a Baby in Queensland Survey 2012 was used to compare the experiences of 655 CALD women to those of 4049 non-CALD women in Queensland, Australia, across three stages of maternity care: pregnancy, labour and birth, and after birth. After adjustment for model of maternity care received and socio-demographic characteristics, CALD women were significantly more likely than non-CALD women to experience suboptimal staff technical competence in pregnancy, overall perceived safety in pregnancy and labour/birth, and interpersonal sensitivity in pregnancy and labour/birth. Approximately 50% of CALD women did not have the choice to use a translator or interpreter, or the gender of their care provider, during labour and birth. Thirteen themes of preferences and needs of CALD maternity care consumers based on ethnicity, cultural beliefs, or traditions were identified, however, these were rarely met. Findings imply that CALD women in Queensland experience disadvantageous maternity care with regards to perceived staff technical competence, safety, and interpersonal sensitivity, and receive care that lacks cultural competence. Improved access to support persons, continuity and choice of carer, and staff availability and training is recommended.

Investigating in-domain data requirements for PLDA training

This paper analyzes the limitations upon the amount of in- domain (NIST SREs) data required for training a probabilistic linear discriminant analysis (PLDA) speaker verification system based on out-domain (Switchboard) total variability subspaces. By limiting the number of speakers, the number of sessions per speaker and the length of active speech per session available in the target domain for PLDA training, we investigated the relative effect of these three parameters on PLDA speaker verification performance in the NIST 2008 and NIST 2010 speaker recognition evaluation datasets. Experimental results indicate that while these parameters depend highly on each other, to beat out-domain PLDA training, more than 10 seconds of active speech should be available for at least 4 sessions/speaker for a minimum of 800 speakers. If further data is available, considerable improvement can be made over solely out-domain PLDA training.

A virtual radiation therapy workflow training simulation

Aim Simulation forms an increasingly vital component of clinical skills development in a wide range of professional disciplines. Simulation of clinical techniques and equipment is designed to better prepare students for placement by providing an opportunity to learn technical skills in a “safe” academic environment. In radiotherapy training over the last decade or so this has predominantly comprised treatment planning software and small ancillary equipment such as mould room apparatus. Recent virtual reality developments have dramatically changed this approach. Innovative new simulation applications and file processing and interrogation software have helped to fill in the gaps to provide a streamlined virtual workflow solution. This paper outlines the innovations that have enabled this, along with an evaluation of the impact on students and educators. Method Virtual reality software and workflow applications have been developed to enable the following steps of radiation therapy to be simulated in an academic environment: CT scanning using a 3D virtual CT scanner simulation; batch CT duplication; treatment planning; 3D plan evaluation using a virtual linear accelerator; quantitative plan assessment, patient setup with lasers; and image guided radiotherapy software. Results Evaluation of the impact of the virtual reality workflow system highlighted substantial time saving for academic staff as well as positive feedback from students relating to preparation for clinical placements. Students valued practice in the “safe” environment and the opportunity to understand the clinical workflow ahead of clinical department experience. Conclusion Simulation of most of the radiation therapy workflow and tasks is feasible using a raft of virtual reality simulation applications and supporting software. Benefits of this approach include time-saving, embedding of a case-study based approach, increased student confidence, and optimal use of the clinical environment. Ongoing work seeks to determine the impact of simulation on clinical skills.

Review and evaluation of mindfulness-based iPhone apps

Background There is growing evidence for the positive impact of mindfulness on wellbeing. Mindfulness-based mobile apps may have potential as an alternative delivery medium for training. While there are hundreds of such apps, there is little information on their quality. Objective This study aimed to conduct a systematic review of mindfulness-based iPhone mobile apps and to evaluate their quality using a recently-developed expert rating scale, the Mobile Application Rating Scale (MARS). It also aimed to describe features of selected high-quality mindfulness apps. Methods A search for “mindfulness” was conducted in iTunes and Google Apps Marketplace. Apps that provided mindfulness training and education were included. Those containing only reminders, timers or guided meditation tracks were excluded. An expert rater reviewed and rated app quality using the MARS engagement, functionality, visual aesthetics, information quality and subjective quality subscales. A second rater provided MARS ratings on 30% of the apps for inter-rater reliability purposes. Results The “mindfulness” search identified 700 apps. However, 94 were duplicates, 6 were not accessible and 40 were not in English. Of the remaining 560, 23 apps met inclusion criteria and were reviewed. The median MARS score was 3.2 (out of 5.0), which exceeded the minimum acceptable score (3.0). The Headspace app had the highest average score (4.0), followed by Smiling Mind (3.7), iMindfulness (3.5) and Mindfulness Daily (3.5). There was a high level of inter-rater reliability between the two MARS raters. Conclusions Though many apps claim to be mindfulness-related, most were guided meditation apps, timers, or reminders. Very few had high ratings on the MARS subscales of visual aesthetics, engagement, functionality or information quality. Little evidence is available on the efficacy of the apps in developing mindfulness.

Evaluation of an intervention for patients with alcohol-related injuries: results of a mixed methods study

Objective: To explore the effect of education and training on the delivery of alcohol screening and brief intervention and referral to high-risk patients in a hospital setting. Main outcome measures included; delivery of training; practice change in relation to staff performing alcohol screening, brief intervention and referrals. Methods: Observational study design using mixed methods set in a tertiary referral hospital. Pre-post assessment of medical records and semi-structured interviews with key informants. Results: Routine screening for substance misuse (9% pre / 71.4% post) and wellbeing concerns (6.6% pre / 15 % post) was more frequent following the introduction of resources and staff participation in educational workshops. There was no evidence of a concomitant increase in delivery of brief intervention or referrals to services. Implementation challenges, including time constraints and staff attitudes, and enablers such as collaboration and visible pathways, were identified. Conclusion: Rates of patient screening increased, however barriers to delivery of brief intervention and referrals remained. Implementation strategies targeting specific barriers and enablers to introducing interventions are both required to improve the application of secondary prevention for patients in acute settings. Implications: Educational training, formalised liaison between services, systematised early intervention protocols, and continuous quality improvement processes will progress service delivery in this area.

Evaluation of a Drunk Driving Warning System

The Drunk Driving Warning System is an alcohol interlock based on performance of the Critical Tracking Task (CTT). An evaluation was undertaken to determine CTT sensitivity to blood alcohol concentration (BAC), particularly at .05 g/100 ml. Subjects were 36 males in 3 age groups (18, 21 to 25, 35 years and above) divided into 2 alcohol consumption categories ("light" and "heavy"), and scored on 4 training and 2 test days (one alcohol and one placebo). The CTT performance declined as BAC increased and was significantly impaired at .05 BAC. However, performance was too variable for in-vehicle use. Age and alcohol consumption pattern were without effect.

Development and evaluation of a spatial decision support system for malaria elimination in Bhutan

Background: Bhutan has reduced its malaria incidence significantly in the last 5 years, and is aiming for malaria elimination by 2016. To assist with the management of the Bhutanese malaria elimination programme a spatial decision support system (SDSS) was developed. The current study aims to describe SDSS development and evaluate SDSS utility and acceptability through informant interviews. Methods: The SDSS was developed based on the open-source Quantum geographical information system (QGIS) and piloted to support the distribution of long-lasting insecticidal nets (LLINs) and indoor residual spraying (IRS) in the two sub-districts of Samdrup Jongkhar District. It was subsequently used to support reactive case detection (RACD) in the two sub-districts of Samdrup Jongkhar and two additional sub-districts in Sarpang District. Interviews were conducted to ascertain perceptions on utility and acceptability of 11 informants using the SDSS, including programme and district managers, and field workers. Results: A total of 1502 households with a population of 7165 were enumerated in the four sub-districts, and a total of 3491 LLINs were distributed with one LLIN per 1.7 persons. A total of 279 households representing 728 residents were involved with RACD. Informants considered that the SDSS was an improvement on previous methods for organizing LLIN distribution, IRS and RACD, and could be easily integrated into routine malaria and other vector-borne disease surveillance systems. Informants identified some challenges at the programme and field level, including the need for more skilled personnel to manage the SDSS, and more training to improve the effectiveness of SDSS implementation and use of hardware. Conclusions: The SDSS was well accepted and informants expected its use to be extended to other malaria reporting districts and other vector-borne diseases. Challenges associated with efficient SDSS use included adequate skills and knowledge, access to training and support, and availability of hardware including computers and global positioning system receivers.

Evaluation of a 7-gene genetic profile for athletic endurance phenotype in Ironman championship triathletes

Polygenic profiling has been proposed for elite endurance performance, using an additive model determining the proportion of optimal alleles in endurance athletes. To investigate this model’s utility for elite triathletes, we genotyped seven polymorphisms previously associated with an endurance polygenic profile (ACE Ins/Del, ACTN3 Arg577Ter, AMPD1 Gln12Ter, CKMM 1170bp/985+185bp, HFE His63Asp, GDF8 Lys153Arg and PPARGC1A Gly482Ser) in a cohort of 196 elite athletes who participated in the 2008 Kona Ironman championship triathlon. Mean performance time (PT) was not significantly different in individual marker analysis. Age, sex, and continent of origin had a significant influence on PT and were adjusted for. Only the AMPD1 endurance-optimal Gln allele was found to be significantly associated with an improvement in PT (model p=5.79 x 10-17, AMPD1 genotype p=0.01). Individual genotypes were combined into a total genotype score (TGS); TGS distribution ranged from 28.6 to 92.9, concordant with prior studies in endurance athletes (mean±SD: 60.75±12.95). TGS distribution was shifted toward higher TGS in the top 10% of athletes, though the mean TGS was not significantly different (p=0.164) and not significantly associated with PT even when adjusted for age, sex, and origin. Receiver operating characteristic curve analysis determined that TGS alone could not significantly predict athlete finishing time with discriminating sensitivity and specificity for three outcomes (less than median PT, less than mean PT, or in the top 10%), though models with the age, sex, continent of origin, and either TGS or AMPD1 genotype could. These results suggest three things: that more sophisticated genetic models may be necessary to accurately predict athlete finishing time in endurance events; that non-genetic factors such as training are hugely influential and should be included in genetic analyses to prevent confounding; and that large collaborations may be necessary to obtain sufficient sample sizes for powerful and complex analyses of endurance performance.

In-training assessment - Its potential in enhancing clinical teaching

Context In-training assessment (ITA) has established its place alongside formative and summative assessment at both the undergraduate and postgraduate level. In this paper the authors aimed to identify those characteristics of ITA that could enhance clinical teaching. Methods A literature review and discussions by an expert working group at the Ninth Cambridge Conference identified the aspects of ITA that could enhance clinical teaching. Results The features of ITA identified included defining the specific benefits to the learner, teacher and institution, and highlighting the patient as the context for ITA and clinical teaching. The ‘mapping’ of a learner’s progress towards the clinical teaching objectives by using multiple assessments over time, by multiple observers in both a systematic and opportunistic way correlates with the incremental nature of reaching clinical competence. Conclusions The importance of ITA based on both direct and indirect evidence of what the learner actually does in the real clinical setting is emphasized. Particular attention is given to addressing concerns in the more controversial areas of assessor training, ratings and documentation for ITA. Areas for future research are also identified.

Training for reduction of design waste

Construction and demolition (C&D) waste have negative impacts on the environment. As a significant proportion of C&D waste is related to the design stage of a project, there is an opportunity for architects to reduce the waste. However, research suggests that many architects often do not understand the impact of their design on waste generation. Training and education are proposed by current researchers to improve architects’ knowledge; however, this has not been adequately validated as a viable approach to solving waste issues. This research investigates architects’ perceptions towards waste management in the design phase, and determines whether they feel they are adequately skilled in reducing C&D waste. Questionnaire surveys were distributed to architects from 98 architectural firms and 25 completed surveys were returned. The results show that while architects are aware of the relationship between design and waste, ‘extra time’ and ‘lack of knowledge’ are the key barriers to implementing waste reduction strategies. In addition, the majority of respondents acknowledge their lack of skill to reduce waste through design evaluation. Therefore, training programmes can be a viable strategy to enable them to address the pressing issue of C&D waste reduction.

Texture for Script identification

The problem of determining the script and language of a document image has a number of important applications in the field of document analysis, such as indexing and sorting of large collections of such images, or as a precursor to optical character recognition (OCR). In this paper, we investigate the use of texture as a tool for determining the script of a document image, based on the observation that text has a distinct visual texture. An experimental evaluation of a number of commonly used texture features is conducted on a newly created script database, providing a qualitative measure of which features are most appropriate for this task. Strategies for improving classification results in situations with limited training data and multiple font types are also proposed.