Maddoxsosa5112

This study explores the use of deep learning-based methods for the automatic detection of COVID-19. Specifically, we aim to investigate the involvement of the virus in the respiratory system by analysing breathing and coughing sounds. Our hypothesis resides in the complementarity of both data types for the task at hand. Therefore, we focus on the analysis of fusion mechanisms to enrich the information available for the diagnosis. In this work, we introduce a novel injection fusion mechanism that considers the embedded representations learned from one data type to extract the embedded representations of the other data type. Our experiments are performed on a crowdsourced database with breathing and coughing sounds recorded using both a web-based application, and a smartphone app. The results obtained support the feasibility of the injection fusion mechanism presented, as the models trained with this mechanism outperform single-type models and multi-type models using conventional fusion mechanisms.Cognitive Behavioral Therapy (CBT) is a goal-oriented psychotherapy for mental health concerns implemented in a conversational setting. The quality of a CBT session is typically assessed by trained human raters who manually assign pre-defined session-level behavioral codes. In this paper, we develop an end-to-end pipeline that converts speech audio to diarized and transcribed text and extracts linguistic features to code the CBT sessions automatically. We investigate both word-level and utterance-level features and propose feature fusion strategies to combine them. The utterance level features include dialog act tags as well as behavioral codes drawn from another well-known talk psychotherapy called Motivational Interviewing (MI). We propose a novel method to augment the word-based features with the utterance level tags for subsequent CBT code estimation. Experiments show that our new fusion strategy outperforms all the studied features, both when used individually and when fused by direct concatenation. see more We also find that incorporating a sentence segmentation module can further improve the overall system given the preponderance of multi-utterance conversational turns in CBT sessions.Management of type 1 diabetes (T1D) requires affected individuals to perform multiple daily actions to keep their blood glucose levels within the safe rage and avoid adverse hypo-/hyperglycemic episodes. Decision support systems (DSS) for T1D are composite tools that implement multiple software modules aiming to ease such a burden and to improve glucose control. At the University of Padova, we are developing a new DSS that currently integrate a smart insulin bolus calculator for optimal insulin dosing and a rescue carbohydrate intake advisor to tackle hypoglycemia. However, a module specifically targeting hyperglycemia, that suggests the administration of corrective insulin boluses (CIB), is still missing. For such a scope, this work aims to assess a recent literature methodology, proposed by Aleppo et al., which provides a simple strategy for dealing with hyperglycemia. The methodology is tested retrospectively on clinical data of individuals with T1D. In particular, here we leveraged a novel in silico tool that first identifies a non-linear model of glucose-insulin dynamics on data, then uses such model to simulate and compare the glucose trace obtained by "replaying" the recorded scenario and the glucose trace obtained using the CIB delivery strategy under evaluation. Results show that the CIB delivery strategy significantly reduce the percentage of time spent in hyperglycemia (-15.63%) without inducing any hypoglycemic episode, demonstrating both safety and efficacy of its use. These preliminary results suggest that the CIB delivery strategy proposed by Aleppo et al. is a promising candidate to be included in our system to counteract hyperglycemia. Future work will extensively evaluate the methodology and will compare it against other competing approaches.People with type 1 diabetes (T1D) need exogenous insulin administrations several times a day. The amount of injected insulin is key for maintaining the concentration of blood glucose (BG) within a physiological safe range. According to well-established clinical guidelines, insulin dosing at mealtime is calculated through an empirical formula which, however, does not take advantage of the knowledge of BG trend provided in real-time by continuous glucose monitoring (CGM) sensors. To overcome suboptimal insulin dosage, we recently used machine learning techniques to build two new models, one linear and one nonlinear, which incorporate BG trend information.In this work, we propose an ensemble learning method for mealtime insulin bolus estimation based on dynamic voting, which combines the two models by taking advantage of where each alternative performs better. Being the resulting model black-box, a tool that enables its interpretability was applied to evaluate the contribution of each feature. The proposed model was trained using a synthetic dataset having information on 100 virtual subjects with different mealtime conditions, and its performance was evaluated within a simulated environment.The benefit given by the ensemble method compared to the single models was confirmed by the high time within the target glycemic range, and the trade-off reached in terms of time spent below and above this range. Moreover, the model interpretation pointed out the key role played by the information on BG dynamics in the estimation of insulin dosage.Image-based tracking of laparoscopic instruments plays a fundamental role in computer and robotic-assisted surgeries by aiding surgeons and increasing patient safety. Computer vision contests, such as the Robust Medical Instrument Segmentation (ROBUST-MIS) Challenge, seek to encourage the development of robust models for such purposes, providing large, diverse, and high-quality datasets. To date, most of the existing models for instance segmentation of medical instruments were based on two-stage detectors, which provide robust results but are nowhere near to the real-time, running at 5 frames-per-second (fps) at most. However, for the method to be clinically applicable, a real-time capability is utmost required along with high accuracy. In this paper, we propose the addition of attention mechanisms to the YOLACT architecture to allow real-time instance segmentation of instruments with improved accuracy on the ROBUST-MIS dataset. Our proposed approach achieves competitive performance compared to the winner of the 2019 ROBUST-MIS challenge in terms of robustness scores, obtaining 0.313 ML_DSC and 0.338 MLNSD while reaching real-time performance at >45 fps.This paper presents a trend analysis of the COVID-19 pandemics in Mexico. The studies were run in a subnational basis because they are more useful that way, providing important information about the pandemic to local authorities. Unlike classic approaches in the literature, the trend analysis presented here is not based on the variations in the number of infections along time, but rather on the predicted value of the final number of infections, which is updated every day employing new data. Results for four states and four cities, selected among the most populated in Mexico, are presented. The model was able to suitably fit the local data for the selected regions under evaluation. Moreover, the trend analysis enabled one to assess the accuracy of the forecasts.Video-based monitoring of patients in the neonatal intensive care unit (NICU) has great potential for improving patient care. Video-based detection of clinical events, such as bottle feeding, would represent a step towards semi-automated charting of clinical events. Recording such events contemporaneously would address the limitations associated with retrospective charting. Such a system could also support oral feeding assessment tools, as the patient's feeding skills and nutrition are pivotal in monitoring their growth. We therefore leverage transfer learning using a pretrained VGG-16 model to classify images obtained during an intervention, to determine if a bottle-feeding event is occurring. Additionally, we explore a data expansion technique by extracting similar-feature images from publicly available sources to supplement our dataset of real NICU patients to address data scarcity. This work also visualizes and quantifies the gap between the source domain (ImageNet data subset) and target domain (NICU dataset), thereby illustrating the impact of expanding our training set for knowledge transfer proficiency. Results show an increase of over 18% in sensitivity after data expansion. Analysis of network activation maps indicates that data expansion is able to reduce the distance between the source and target domains.Diagnosis of temporomandibular joint (TMJ) Osteoarthritis (OA) before serious degradation of cartilage and subchondral bone occurs can help prevent chronic pain and disability. Clinical, radiomic, and protein markers collected from TMJ OA patients have been shown to be predictive of OA onset. Since protein data can often be unavailable for clinical diagnosis, we harnessed the learning using privileged information (LUPI) paradigm to make use of protein markers only during classifier training. Three different LUPI algorithms are compared with traditional machine learning models on a dataset extracted from 92 unique OA patients and controls. The best classifier performance of 0.80 AUC and 75.6 accuracy was obtained from the KRVFL+ model using privileged protein features. Results show that LUPI-based algorithms using privileged protein data can improve final diagnostic performance of TMJ OA classifiers without needing protein microarray data during classifier diagnosis.In emergency medicine, workforce planning needs to satisfy a number of constraints. There are hard constraints regarding qualifications and soft constraints regarding the wishes of the personnel. One instance of such a planning problem is the assignment of lifeguards at the coasts of the North Sea and the Baltic Sea in Germany. These lifeguards are volunteers and thus accounting for wishes is crucial while qualification constraints must be satisfied nevertheless. This paper presents a genetic algorithm that solves this problem with sub-second runtime. We compare this genetic algorithm to a brute force solution creating optimal solutions at the expense of larger runtime complexity. The genetic approach outperforms the brute force approach in terms of runtime when there are more than 3 places of deployment while consistently producing optimal solutions within less than 10 generations.Traditional methods of posture evaluation carried out by physical therapists manually measure or test the alignments of body segments, investing a long time for its development and adding an error percentage related to the level of professional expertise. The present study uses a system of two dimensions photogrammetry to investigate its applicability on measurement of posture parameters and the variation of the measurements using different photographic cameras locate at different distances from the subject. The "marker automatic measurement" system (LAM) filters and segments body markers on photographic images. Data were collected using a semi-professional, a mid-range cellphone and a sports camera. Tests were recorded by placing the camera at 2.50, 2.00 and 1.80 meters from the subject, and the lens at a height of 1.10, 1.00 and 0.97 meters with an illuminance of 29.92 lux. Subsequently, 30 volunteers participated in the postural tests. The Measurements were made on frontal, anterior and posterior planes as well as sagittal plane.