3 research outputs found
Representation learning for improved distance and risk metrics
No full textThesis: M. Eng., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2018.This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.Cataloged from student-submitted PDF version of thesis.Includes bibliographical references (pages 43-49).In this thesis, we present methods in representation learning for time series in two areas: metric learning and risk stratification. We focus on metric learning due to the importance of computing distances between examples in learning algorithms and present Jiffy, a simple and scalable distance metric learning method for multivariate time series. Our approach is to reframe the task as a representation learning problem -- rather than design an elaborate distance function, we use a CNN to learn an embedding such that the Euclidean distance is effective. Experiments on a diverse set of multivariate time series datasets show that our approach consistently outperforms existing methods. We then focus on risk stratification because of its clinical importance in identifying patients at high risk for an adverse outcome. We use segments of a patient's ECG signal to predict that patient's risk of cardiovascular death within 90 days. In contrast to other work, we work directly with the raw ECG signal to learn a representation with predictive power. Our method produces a risk metric for cardiovascular death with state-of-the-art performance when compared to methods that rely on expert-designed representations.by Divya Shanmugam.M. Eng
Constructing the CORD-19 Vaccine Dataset
No full textWe introduce new dataset \u27CORD-19-Vaccination\u27 to cater to scientists specifically looking into COVID-19 vaccine-related research. This dataset is extracted from CORD-19 dataset [Wang et al., 2020] and augmented with new columns for language detail, author demography, keywords, and topic per paper. Facebook\u27s fastText model is used to identify languages [Joulin et al., 2016]. To establish author demography (author affiliation, lab/institution location, and lab/institution country columns) we processed the JSON file for each paper and then further enhanced using Google\u27s search API to determine country values. \u27Yake\u27 was used to extract keywords from the title, abstract, and body of each paper and the LDA (Latent Dirichlet Allocation) algorithm was used to add topic information [Campos et al., 2020, 2018a,b]. To evaluate the dataset, we demonstrate a question-answering task like the one used in the CORD-19 Kaggle challenge [Goldbloom et al., 2022]. For further evaluation, sequential sentence classification was performed on each paper\u27s abstract using the model from Dernoncourt et al. [2016]. We partially hand annotated the training dataset and used a pre-trained BERT-PubMed layer. \u27CORD- 19-Vaccination\u27 contains 30k research papers and can be immensely valuable for NLP research such as text mining, information extraction, and question answering, specific to the domain of COVID-19 vaccine research
Health-related quality of life at 30 days among Indian patients with acute myocardial infarction results from the ACS QUIK trial
No full textBackground:
Despite a high cardiovascular disease burden, data on patient-reported health status outcomes among individuals with cardiovascular disease in India are limited.
Methods and Results:
Between November 2014 and November 2016, we collected health-related quality of life data among 1261 participants in the ACS QUIK trial (Acute Coronary Syndrome Quality Improvement in Kerala). We used a translated, validated version of the Seattle Angina Questionnaire administered 30 days after discharge for acute myocardial infarction, wherein higher scores represent better health status. We compared results across sex, myocardial infarction type, and randomization status using regression models that account for clustering and temporal trends. Mean (SD) age was 60.8 (13.7) years, 62% were men, and 63% presented with ST-segment–elevation myocardial infarction. More than 2 out of 5 respondents (44%) experienced angina 30 days after hospitalization, but most (68% of respondents with angina; 27% of the total sample) experienced it less than once per week (Seattle Angina Questionnaire angina frequency score 60). Respondents rated high median (interquartile range [IQR]) scores for angina frequency (100.0 [80.0–100.0]) overall with similar unadjusted scores by sex, but between-hospitality variability was high. Median (IQR) physical limitation scale response was 58.3 (41.7–77.8), which is consistent with limitations in moderate- and high-intensity activities at 30-day follow-up. Older respondents had more angina frequency and physical limitations and lower treatment satisfaction and quality of life. Women had greater physical limitations (median [IQR], 52.8 [38.9–72.2] for women versus median [IQR], 61.1 [44.4–80.6] for men; P<0.01). Overall treatment satisfaction was high with median (IQR) score, 81.3 (75.0–93.8), but overall quality of life was lower with median (IQR) score, 66.7 (50.0–83.3). Allocation to the quality improvement intervention group had the strongest direct association with higher quality of life (difference, 4.2; P=0.03), but overall effects were modest.
Conclusions:
This study represents the largest report of quality of life among myocardial infarction survivors in India with variability across age, sex, and quality improvement intervention status. Wide variability demonstrated across hospitals warrants further study.
Clinical Trial Registration:
URL: https://www.clinicaltrials.gov. Unique identifier: NCT02256657
