Improving sentiment analysis on clinical narratives by exploiting UMLS semantic types

Sentiments associated with assessments and observations recorded in a clinical narrative can often indicate a patient's health status. To perform sentiment analysis on clinical narratives, domain-specific knowledge concerning meanings of medical terms is required. In this study, semantic types in the Unified Medical Language System (UMLS) are exploited to improve lexicon-based sentiment classification methods. For sentiment classification using SentiWordNet, the overall accuracy is improved from 0.582 to 0.710 by using logistic regression to determine appropriate polarity scores for UMLS 'Disorders' semantic types. For sentiment classification using a trained lexicon, when disorder terms in a training set are replaced with their semantic types, classification accuracies are improved on some data segments containing specific semantic types. To select an appropriate classification method for a given data segment, classifier combination is proposed. Using classifier combination, classification accuracies are improved on most data segments, with the overall accuracy of 0.882 being obtained.

Towards Chinese clinical named entity recognition by dynamic embedding using domain-specific knowledge

The task of electronic medical record named entity recognition (NER) refers to automatically identify all kinds of named entities in the medical record text. Chinese clinical NER remains a major challenge. One of the main reasons is that Chinese word segmentation will lead to the wrong downstream works. Besides, existing methods only use the information of the general field, not consider the knowledge from field of medicine. To address these issues, we propose a dynamic embedding method based on dynamic attention which combines features of both character and word in embedding layer. Domain knowledge is provided by word vector trained by domain dataset. In addition, spatial attention is added to enable the model to obtain more and more effective context encoding information. Finally, we conduct extensive experiments to demonstrate the effectiveness of our proposed algorithm. Experiments on CCKS2017 and Common dataset shows that the proposed method outperforms the baseline.

Cognition and Health Literacy in Older Adults' Recall of Self-Care Information

Purpose of the Study

Health literacy is associated with health outcomes presumably because it influences the understanding of information needed for self-care. However, little is known about the language comprehension mechanisms that underpin health literacy.

Design and Methods

We explored the relationship between a commonly used measure of health literacy (Short Test of Functional Health Literacy in Adults [STOFHLA]) and comprehension of health information among 145 older adults.

Results

Results showed that performance on the STOFHLA was associated with recall of health information. Consistent with the Process-Knowledge Model of Health Literacy, mediation analysis showed that both processing capacity and knowledge mediated the association between health literacy and recall of health information. In addition, knowledge moderated the effects of processing capacity limits, such that processing capacity was less likely to be associated with recall for older adults with higher levels of knowledge.

Implications

These findings suggest that knowledge contributes to health literacy and can compensate for deficits in processing capacity to support comprehension of health information among older adults. The implications of these findings for improving patient education materials for older adults with inadequate health literacy are discussed.

The impact of domain-specific beliefs on decisions and causal judgments

Extensive evidence suggests that people often rely on their causal beliefs in their decisions and causal judgments. To date, however, there is a dearth of research comparing the impact of causal beliefs in different domains. We conducted two experiments to map the influence of domain-specific causal beliefs on the evaluation of empirical evidence when making decisions and subsequent causal judgments. Participants made 120 decisions in a two-alternative forced-choice task, framed in either a medical or a financial domain. Before each decision, participants could actively search for information about the outcome ("occurrence of a disease" or "decrease in a company's share price") on the basis of four cues. To analyze the strength of causal beliefs, we set two cues to have a generative relation to the outcome and two to have a preventive relation to the outcome. To examine the influence of empirical evidence, we manipulated the predictive power (i.e., cue validities) of the cues. Both experiments included a validity switch, where the four selectable cues switched from high to low validity or vice versa. Participants had to make a causal judgment about each cue before and after the validity switch. In the medical domain, participants stuck to the causal information in causal judgments, even when evidence was contradictory, while decisions showed an effect of both empirical and causal information. In contrast, in the financial domain, participants mainly adapted their decisions and judgments to the cue validities. We conclude that the strength of causal beliefs (1) is shaped by the domain, and (2) has a differential influence on the degree to which empirical evidence is taken into account in causal judgments and decision making.