Machine-learning assisted swallowing assessment: a deep learning-based quality improvement tool to screen for post-stroke dysphagia.

Introduction: Post-stroke dysphagia is common and associated with significant morbidity and mortality, rendering bedside screening of significant clinical importance. Using voice as a biomarker coupled with deep learning has the potential to improve patient access to screening and mitigate the subjectivity associated with detecting voice change, a component of several validated screening protocols. Methods: In this single-center study, we developed a proof-of-concept model for automated dysphagia screening and evaluated the performance of this model on training and testing cohorts. Patients were admitted to a comprehensive stroke center, where primary English speakers could follow commands without significant aphasia and participated on a rolling basis. The primary outcome was classification either as a pass or fail equivalent using a dysphagia screening test as a label. Voice data was recorded from patients who spoke a standardized set of vowels, words, and sentences from the National Institute of Health Stroke Scale. Seventy patients were recruited and 68 were included in the analysis, with 40 in training and 28 in testing cohorts, respectively. Speech from patients was segmented into 1,579 audio clips, from which 6,655 Mel-spectrogram images were computed and used as inputs for deep-learning models (DenseNet and ConvNext, separately and together). Clip-level and participant-level swallowing status predictions were obtained through a voting method. Results: The models demonstrated clip-level dysphagia screening sensitivity of 71% and specificity of 77% (F1 = 0.73, AUC = 0.80 [95% CI: 0.78-0.82]). At the participant level, the sensitivity and specificity were 89 and 79%, respectively (F1 = 0.81, AUC = 0.91 [95% CI: 0.77-1.05]). Discussion: This study is the first to demonstrate the feasibility of applying deep learning to classify vocalizations to detect post-stroke dysphagia. Our findings suggest potential for enhancing dysphagia screening in clinical settings. https://github.com/UofTNeurology/masa-open-source.

Investigation of independent reinforcement learning algorithms in multi-agent environments.

Independent reinforcement learning algorithms have no theoretical guarantees for finding the best policy in multi-agent settings. However, in practice, prior works have reported good performance with independent algorithms in some domains and bad performance in others. Moreover, a comprehensive study of the strengths and weaknesses of independent algorithms is lacking in the literature. In this paper, we carry out an empirical comparison of the performance of independent algorithms on seven PettingZoo environments that span the three main categories of multi-agent environments, i.e., cooperative, competitive, and mixed. For the cooperative setting, we show that independent algorithms can perform on par with multi-agent algorithms in fully-observable environments, while adding recurrence improves the learning of independent algorithms in partially-observable environments. In the competitive setting, independent algorithms can perform on par or better than multi-agent algorithms, even in more challenging environments. We also show that agents trained via independent algorithms learn to perform well individually, but fail to learn to cooperate with allies and compete with enemies in mixed environments.

Analysis of Language Embeddings for Classification of Unstructured Pathology Reports.

A pathology report is one of the most significant medical documents providing interpretive insights into the visual appearance of the patient's biopsy sample. In digital pathology, high-resolution images of tissue samples are stored along with pathology reports. Despite the valuable information that pathology reports hold, they are not used in any systematic manner to promote computational pathology. In this work, we focus on analyzing the reports, which are generally unstructured documents written in English with sophisticated and highly specialized medical terminology. We provide a comparative analysis of various embedding models like BioBERT, Clinical BioBERT, BioMed-RoBERTa and Term Frequency-Inverse Document Frequency (TF-IDF), a traditional NLP technique, as well as the combination of embeddings from pre-trained models with TF-IDF. Our results demonstrate the effectiveness of various word embedding techniques for pathology reports.

Supervision and Source Domain Impact on Representation Learning: A Histopathology Case Study.

As many algorithms depend on a suitable representation of data, learning unique features is considered a crucial task. Although supervised techniques using deep neural networks have boosted the performance of representation learning, the need for a large sets of labeled data limits the application of such methods. As an example, high-quality delineations of regions of interest in the field of pathology is a tedious and time-consuming task due to the large image dimensions. In this work, we explored the performance of a deep neural network and triplet loss in the area of representation learning. We investigated the notion of similarity and dissimilarity in pathology whole-slide images and compared different setups from unsupervised and semi-supervised to supervised learning in our experiments. Additionally, different approaches were tested, applying few-shot learning on two publicly available pathology image datasets. We achieved high accuracy and generalization when the learned representations were applied to two different pathology datasets.

Extracorporeal membrane oxygenation support improves survival of patients with Hantavirus cardiopulmonary syndrome refractory to medical treatment.

OBJECTIVE: The aim of the study was to evaluate the outcome of extracorporeal membrane oxygenation (ECMO) support in Hantavirus cardiopulmonary syndrome (HCPS) patients with a predicted mortality of 100%, and the complications associated with this treatment modality and different cannulation techniques. METHODS: A total of 51 patients with refractory HCPS were supported with ECMO between April 1994 and June 2010. They were divided into group A consisting of the 26 patients treated between 1994 and 2000 and group B consisting of 25 patients treated between 2003 and 2010. No patients were treated between September 2000 and December 2003. Patients in group A were intubated when they became hypoxic and placed on ECMO when they became hemodynamically unstable, whereas patients in group B had elective insertion of vascular sheaths and were almost concurrently intubated and placed on ECMO when they decompensated. Cannulation of the femoral vessels was performed percutaneously in 18 (35.3%) patients and with an open technique in 33 (64.7%) patients. RESULTS: Complications from percutaneous cannulation occurred in 4/18 (22.27%) patients: retroperitoneal hematoma in 2/18 (11.1%) and lower-extremity ischemia in 2/18 (11.1%) patients. Complications from open femoral cannulation occurred in 12/33 (36.3%) patients: bleeding in 10/33 (30.3%) patients and ischemia in 2/33 (6.1%) patients. The overall survival was 66.6% (34 of 51 patients); 56% (14/26) for group A and 80% (20/25) for group B (p = 0.048). There was no difference in mortality regarding the method of cannulation. A trend toward increased mortality in patients with cannulation complications was recognized, but it was not statistically significant. Mortality was not associated with ECMO duration (average 121.7h, range: 5-276h). All survivors recovered and were discharged from the hospital after a mean hospital stay of 19.8 days (range: 10-39 days). CONCLUSIONS: Two-thirds of 51 HCPS patients with a predicted mortality of 100%, who were supported with ECMO, survived and recovered completely. Survival was significantly higher in the second half of the study. Complications associated with both types of femoral cannulation were associated with a trend toward decreased survival, which was not significant.

Extracorporeal membrane oxygenation support improves survival of patients with severe Hantavirus cardiopulmonary syndrome.

OBJECTIVE: The purposes of this study are to evaluate the outcome of extracorporeal membrane oxygenation support in a subgroup of patients with Hantavirus cardiopulmonary syndrome who had a predicted mortality of 100% and to assess the complications associated with this treatment modality and with different cannulation techniques. METHODS: Thirty-eight patients with severe Hantavirus cardiopulmonary syndrome were supported with extracorporeal membrane oxygenation between April 1994 and June 2006. Cannulation of the femoral vessels was performed on an emergency basis by a percutaneous approach in 15 (39.5%) and by an open technique in 23 (60.5%) patients. Duration of extracorporeal membrane oxygenation averaged 132 hours (range: 5-276 hours). RESULTS: Complications from percutaneous cannulation occurred in 4 (26.6%) of 15 patients: retroperitoneal hematoma in 2 (13.3%) and lower extremity ischemia in 2 (13.3%) patients, which resolved after insertion of a distal perfusion cannula. Complications from open femoral cannulation occurred in 8 (34.8%) of 23 patients: severe bleeding in 7 (30.4%) patients and lower extremity ischemia in 1 (4.3%) patient who required a leg amputation. The overall survival was 60.5% (23/38 patients). Six (40%) of the 15 patients cannulated percutaneously and 9 (39.1%) of 23 patients who had open cannulation died. All survivors recovered completely and were discharged from the hospital after a mean hospital stay of 20.8 days (range: 10-39 days). CONCLUSIONS: Almost two thirds of the patients with severe Hantavirus cardiopulmonary syndrome who were supported with extracorporeal circulation survived and recovered completely. The complications associated with both types of femoral cannulation may be attributed to the fact that all patients were in shock or in full cardiac arrest, and the procedure had to be done expeditiously. Earlier institution of extracorporeal membrane oxygenation may decrease the complication rates and improve the overall survival.

Case report: three farmworkers who gave birth to infants with birth defects closely grouped in time and place-Florida and North Carolina, 2004-2005.

CONTEXT: There is little evidence linking adverse reproductive effects to exposure to specific pesticides during pregnancy. CASE PRESENTATION: In February 2005, three infants with congenital anomalies were identified in Collier County, Florida, who were born within 8 weeks of one another and whose mothers worked for the same tomato grower. The mothers worked on the grower's Florida farms in 2004 before transferring to its North Carolina farms. All three worked during the period of organogenesis in fields recently treated with several pesticides. The Florida and North Carolina farms were inspected by regulatory agencies, and in each state a large number of violations were identified and record fines were levied. DISCUSSION: Despite the suggestive evidence, a causal link could not be established between pesticide exposures and the birth defects in the three infants. Nonetheless, the prenatal pesticide exposures experienced by the mothers of the three infants is cause for concern. Farmworkers need greater protections against pesticides. These include increased efforts to publicize and comply with both the U.S. Environmental Protections Agency's Worker Protection Standard and pesticide label requirements, enhanced procedures to ensure pesticide applicator competency, and recommendations to growers to adopt work practices to reduce pesticide exposures. RELEVANCE TO PROFESSIONAL PRACTICE: The findings from this report reinforce the need to reduce pesticide exposures among farmworkers. In addition, they support the need for epidemiologic studies to examine the role of pesticide exposure in the etiology of congenital anomalies.

Hantavirus cardiopulmonary syndrome.

Hantavirus cardiopulmonary syndrome (HCPS) is caused by infection with multiple types of hantaviruses throughout the Americas. All strains induce a pulmonary capillary leak syndrome with cardiogenic shock in severe cases, and almost all strains have an overall mortality rate of 35%. Some strains in South America are commonly associated with either mild disease without pulmonary edema or severe disease with pulmonary hemorrhage. Early recognition during the thrombocytopenic prodrome phase and transport to intensive care improve survival, due to very rapid progression of respiratory failure and shock in some patients. Limited fluid replacement, early inotropic therapy, and mechanical ventilation also improve survival. Extracorporeal membrane oxygenation (ECMO) has been life saving for patients with refractory shock. Ribavirin does not appear to reduce mortality, but hyperimmune serum offers a promising future therapy because survival is correlated with higher neutralizing antibody titers at admission.

Diagnosis and treatment of new world hantavirus infections.

PURPOSE OF REVIEW: The purpose of this review is to summarize the current knowledge regarding the diagnosis and treatment of indigenous new world hantavirus infections. RECENT FINDINGS: Recent studies have defined the incubation period of new world hantavirus infections, provided additional evidence for person-to-person transmission of Andes virus, described a rapid method for the presumptive diagnosis of infection in the cardiopulmonary phase through a review of the peripheral smear, and suggested that intravenous ribavirin is probably not effective for the treatment of new world hantavirus infections when started in the cardiopulmonary phase. SUMMARY: Presumptive diagnosis may be made by a review of the peripheral blood smear after the onset of the cardiopulmonary phase. Critical care management includes the avoidance of fluid overload, pressors to maintain cardiac output, and the use of extracorporeal membrane oxygenation in the most severe cases, but treatment with intravenous ribavirin is probably not effective.