Purpose
User-generated content on social media sites, such as health-related online forums, offers researchers a tantalizing amount of information, but concerns regarding scientific application of such data remain. This paper compares and contrasts symptom cluster patterns derived from messages on a breast cancer forum with those from a symptom checklist completed by breast cancer survivors participating in a research study.
Methods
Over 50,000 messages generated by 12,991 users of the breast cancer forum on MedHelp.org were transformed into a standard form and examined for the co-occurrence of 25 symptoms. The k-medoid clustering method was used to determine appropriate placement of symptoms within clusters. Findings were compared with a similar analysis of a symptom checklist administered to 653 breast cancer survivors participating in a research study.
Results
The following clusters were identified using forum data: menopausal/psychological, pain/fatigue, gastrointestinal, and miscellaneous. Study data generated the clusters: menopausal, pain, fatigue/sleep/gastrointestinal, psychological, and increased weight/appetite. Although the clusters are somewhat different, many symptoms that clustered together in the social media analysis remained together in the analysis of the study participants. Density of connections between symptoms, as reflected by rates of co-occurrence and similarity, was higher in the study data.
Conclusions
The copious amount of data generated by social media outlets can augment findings from traditional data sources. When different sources of information are combined, areas of overlap and discrepancy can be detected, perhaps giving researchers a more accurate picture of reality. However, data derived from social media must be used carefully and with understanding of its limitations.
This study provides evidence of data set mentions and citations in multiple disciplines based on a content analysis of 600 publications in PLoS One. We find that data set mentions and citations varied greatly among disciplines in terms of how data sets were collected, referenced, and curated. While a majority of articles provided free access to data, formal ways of data attribution such as DOIs and data citations were used in a limited number of articles. In addition, data reuse took place in less than 30% of the publications that used data, suggesting that researchers are still inclined to create and use their own data sets, rather than reusing previously curated data. This paper provides a comprehensive understanding of how data sets are used in science and helps institutions and publishers make useful data policies.
We investigate the quality of task specific word embeddings created with relatively small, targeted corpora. We present a comprehensive evaluation framework including both intrinsic and extrinsic evaluation that can be expanded to named entities beyond drug name. Intrinsic evaluation results tell that drug name embeddings created with a domain specific document corpus outperformed the previously published versions that derived from a very large general text corpus. Extrinsic evaluation uses word embedding for the task of drug name recognition with Bi-LSTM model and the results demonstrate the advantage of using domain-specific word embeddings as the only input feature for drug name recognition with F1-score achieving 0.91. This work suggests that it may be advantageous to derive domain specific embeddings for certain tasks even when the domain specific corpus is of limited size.
BackgroundDue to the high cost and low success rate in new drug development, systematic drug repositioning methods are exploited to find new indications for existing drugs.ObjectiveWe sought to propose a new computational drug repositioning method to identify repositioning drugs for Parkinson disease (PD).MethodsWe developed a novel heterogeneous network mining repositioning method that constructed a 3-layer network of disease, drug, and adverse drug reaction and involved user-generated data from online health communities to identify potential candidate drugs for PD.ResultsWe identified 44 non-Parkinson drugs by using the proposed approach, with data collected from both pharmaceutical databases and online health communities. Based on the further literature analysis, we found literature evidence for 28 drugs.ConclusionsIn summary, the proposed heterogeneous network mining repositioning approach is promising for identifying repositioning candidates for PD. It shows that adverse drug reactions are potential intermediaries to reveal relationships between disease and drug.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.