Computer-aided autism diagnosis using visual attention models and eye-tracking: replication and improvement proposal.

BACKGROUND: Autism Spectrum Disorder (ASD) diagnosis can be aided by approaches based on eye-tracking signals. Recently, the feasibility of building Visual Attention Models (VAMs) from features extracted from visual stimuli and their use for classifying cases and controls has been demonstrated using Neural Networks and Support Vector Machines. The present work has three aims: 1) to evaluate whether the trained classifier from the previous study was generalist enough to classify new samples with a new stimulus; 2) to replicate the previously approach to train a new classifier with a new dataset; 3) to evaluate the performance of classifiers obtained by a new classification algorithm (Random Forest) using the previous and the current datasets. METHODS: The previously approach was replicated with a new stimulus and new sample, 44 from the Typical Development group and 33 from the ASD group. After the replication, Random Forest classifier was tested to substitute Neural Networks algorithm. RESULTS: The test with the trained classifier reached an AUC of 0.56, suggesting that the trained classifier requires retraining of the VAMs when changing the stimulus. The replication results reached an AUC of 0.71, indicating the potential of generalization of the approach for aiding ASD diagnosis, as long as the stimulus is similar to the originally proposed. The results achieved with Random Forest were superior to those achieved with the original approach, with an average AUC of 0.95 for the previous dataset and 0.74 for the new dataset. CONCLUSION: In summary, the results of the replication experiment were satisfactory, which suggests the robustness of the approach and the VAM-based approaches feasibility to aid in ASD diagnosis. The proposed method change improved the classification performance. Some limitations are discussed and additional studies are encouraged to test other conditions and scenarios.

Computer-aided autism diagnosis based on visual attention models using eye tracking.

An advantage of using eye tracking for diagnosis is that it is non-invasive and can be performed in individuals with different functional levels and ages. Computer/aided diagnosis using eye tracking data is commonly based on eye fixation points in some regions of interest (ROI) in an image. However, besides the need for every ROI demarcation in each image or video frame used in the experiment, the diversity of visual features contained in each ROI may compromise the characterization of visual attention in each group (case or control) and consequent diagnosis accuracy. Although some approaches use eye tracking signals for aiding diagnosis, it is still a challenge to identify frames of interest when videos are used as stimuli and to select relevant characteristics extracted from the videos. This is mainly observed in applications for autism spectrum disorder (ASD) diagnosis. To address these issues, the present paper proposes: (1) a computational method, integrating concepts of Visual Attention Model, Image Processing and Artificial Intelligence techniques for learning a model for each group (case and control) using eye tracking data, and (2) a supervised classifier that, using the learned models, performs the diagnosis. Although this approach is not disorder-specific, it was tested in the context of ASD diagnosis, obtaining an average of precision, recall and specificity of 90%, 69% and 93%, respectively.

Urban wild meat consumption and trade in central Amazonia.

The switch from hunting wild meat for home consumption to supplying more lucrative city markets in Amazonia can adversely affect some game species. Despite this, information on the amounts of wild meat eaten in Amazonian cities is still limited. We estimated wild meat consumption rates in 5 cities in the State of Amazonas in Brazil through 1046 door-to-door household interviews conducted from 2004 to 2012. With these data, we modeled the relationship between wild meat use and a selection of socioeconomic indices. We then scaled up our model to determine the amounts of wild meat likely to be consumed annually in the 62 urban centers in central Amazonia. A total of 80.3% of all interviewees reported consuming wild meat during an average of 29.3 (CI 11.6) days per year. Most wild meat was reported as bought in local markets (80.1%) or hunted by a family member (14.9%). Twenty-one taxa were cited as consumed, mostly mammals (71.6%), followed by reptiles (23.2%) and then birds (5.2%). The declared frequency of wild meat consumption was positively correlated with the proportion of rural population as well as with the per capita gross domestic product of the municipality (administrative divisions) where the cities were seated. We estimated that as much as 10,691 t of wild meat might be consumed annually in the 62 urban centers within central Amazonia, the equivalent of 6.49 kg per person per year. In monetary terms, this amounts to US$21.72 per person per year or US$35.1 million overall, the latter figure is comparable to fish and timber production in the region. Given this magnitude of wild meat trade in central Amazonia, it is fundamental to integrate this activity into the formal economy and actively develop policies that allow the trade of more resilient taxa and restrict trade in species sensitive to hunting.

Mercado y Consumo Urbano de Carne Silvestre en la Amazonia Central Resumen El cambio de la caza de animales silvestres para consumo del hogar a caza para surtir mercados urbanos más lucrativos en la Amazonia puede afectar negativamente a algunas especies de caza. A pesar de esto, la información sobre la cantidad de carne silvestre que se consume en las ciudades de la Amazonia todavía es limitada. Estimamos las tasas de consumo de carne silvestre en cinco ciudades del Estado de Amazonas, Brasil, por medio de 1046 entrevistas presenciales a hogares realizadas entre 2004 y 2012. Con estos datos modelamos la relación entre el uso de la carne silvestre y una selección de índices socioeconómicos. Después aumentamos nuestro modelo para determinar la cantidad de carne silvestre que tal vez se consume anualmente en los 62 centros urbanos de la Amazonia central. Un total de 80.3% de todos los entrevistados reportaron el consumo de carne silvestre durante un promedio de 29.3 (CI 11.6) días por año. La mayoría de la carne silvestre fue reportada como comprada en mercados locales (80.1%) o cazada por algún integrante de la familia (14.9%). Se citaron 21 taxones como parte del consumo, principalmente mamíferos (71.6%), seguidos por reptiles (23.2%) y aves (5.2%). La frecuencia declarada de consumo de carne silvestre estuvo correlacionada positivamente con la proporción de población rural, así como con el producto doméstico bruto per cápita de la municipalidad (divisiones administrativas) en donde se encuentran las ciudades. Estimamos que un máximo de 10,691 toneladas de carne silvestre podrían ser consumidas anualmente en los 62 centros urbanos de la Amazonia central, el equivalente a 6.49 kg/persona/año. En términos monetarios, esto equivale a US$21.72/persona/año o US$35.1 millones en general. Esta última cifra es comparable con la producción de madera y de peces en la región. Dada esta magnitud del mercado de carne silvestre en la Amazonia central, es fundamental la integración de esta actividad en la economía formal y el desarrollo activo de políticas que permitan el mercado de taxones más resilientes y que restrinjan el mercado para las especies sensibles a la caza.

Major globally disseminated clonal complexes of antimicrobial resistant enterococci associated with infections in cancer patients in Brazil.

Cancer and hematological malignancies constitute major comorbidities in enterococcal infections, but little is known about the characteristics of enterococci affecting cancer patients. The aim of this study was to characterize 132 enterococcal clinical isolates obtained from cancer patients attending a Cancer Reference Center in Brazil between April 2013 and March 2014. Susceptibility to 17 antimicrobial agents was assessed by disk diffusion method. Resistance and virulence genes were investigated by PCR. Multilocus sequence typing (MLST) was performed for selected Enterococcus faecalis and Enterococcus faecium isolates. The predominant species was E. faecalis (108 isolates), followed by E. faecium (18), Enterococcus gallinarum (3), Enterococcus avium (2) and Enterococcus durans (1). Multidrug-resistant (MDR) isolates made up 44.7%, but all isolates were susceptible to fosfomycin, linezolid and glycopeptides. The most prevalent genes associated with erythromycin- and tetracycline-non susceptible isolates were erm(B) (47/71; 66.2%) and tet(M) (24/68; 35.3%), respectively. High-level resistance (HLR) to gentamicin was found in 22 (16.7%) isolates and 13 (59.1%) of them carried the aac(6')-Ie-aph(2″)-Ia gene. HLR to streptomycin was detected in 34 (25.8%) isolates, of which 15 (44.1%) isolates had the ant(6')-Ia gene. The most common virulence genes were gelE (48.9%), esp (30.5%) and asa1 (29.8%). MLST performed for 26 E. faecalis isolates revealed 18 different sequence-types (STs), with seven corresponding to novel STs (625, 626, 627, 628, 629, 630, and 635). On the other hand, nine of 10 E. faecium isolates analyzed by MLST belonged to a single clonal complex, comprised of mostly ST412, which emerged worldwide after mid-2000s, but also two novel STs (963 and 964). We detected major globally disseminated E. faecalis and E. faecium clonal complexes along with novel closely related STs, indicating the fitness and continuous evolution of these hospital-adapted lineages.