Name
Abstract | ||
|---|---|---|
Literature for evidence on factors that put an individual at risk of pressure injury usually focused on identifying independent pressure injury risk factors. It is hard to find how important each factor is through the literature. By extracting casual relations, we can tackle the vast volume of causal knowledge and establish causal graphs. In this paper, we aim to use an unsupervised learning model to extract causal relations between pressure injury and risk factors. The workflow includes data preprocessing, causality determination, causality verification, and knowledge graph drawing. We conduct extensive experiments on a medical literature data set of 10,000 abstracts crawling from Pubmed and compare the knowledge graph we draw with the latest international guideline to verify the accuracy. We study 12 pressure injury risk factors and finally extract 10 relations between pressure injury and risk factors with the correct ratio 8/10, and 17 relations in the risk factor pairs with the correct ratio 16/17. The average credibility of extracting relations between pressure injury and risk factors is 0.7317, and 0.8983 for extracting relations of the 17 risk factor pairs. It indicates that the proposed method of extracting causal relations from the literature of pressure injury has a high degree of credibility. |
Year | DOI | Venue |
|---|---|---|
2020 | 10.1109/ICBK50248.2020.00087 | 2020 IEEE International Conference on Knowledge Graph (ICKG) |
Keywords | DocType | ISBN |
casual extraction,pressure injuries,literature analysis,knowledge graphs | Conference | 978-1-7281-8157-8 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
6 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Siyi Guo | 1 | 0 | 0.34 |
| Liuqi Jin | 2 | 0 | 0.68 |
| Jiaoyun Yang | 3 | 29 | 5.04 |
| Mengyao Jiang | 4 | 0 | 0.34 |
| Lin Han | 5 | 0 | 0.68 |
| Ning An | 6 | 398 | 36.33 |