Table 10 Comparison with existing disease-disease association extraction tools
From: A study on large-scale disease causality discovery from biomedical literature
Tool | Principle | Advantage | Disadvantage | Disease entity | Relation type | Performance |
|---|---|---|---|---|---|---|
Method in this study | A disease causality extraction method through the optimization of the disease causality semantic predicate list | Can achieve a more accurate and flexible disease causality extraction | Disease causality extraction is incomplete; the manual extraction of predicates is time-consuming and labour-intensive and the evaluation of predicates is subjective | Obtain 14,335 standardized disease entities | Use 36 textual predicates and extract 58 types of semantic relationships | Achieve an accuracy of 96.97% in disease causality extraction |
PubTator | A tool supporting biomedical entities and relations search in the biomedical literature, which involves disease association extraction | Improve the model’s ability to generalize to unseen data, has enhanced entity normalization and relation extraction performance | The accuracy of entity annotation and relation extraction remains imperfect, and relations extraction is restricted to abstracts | Involve 12,850 disease entities | Cover 12 types of relations, and two of them express disease relations | F-score is 82%, and demonstrates an overall precision of 90.0% |
SicknessMiner | A deep-learning-driven text-mining tool for disease-disease associations extraction | Provide an easy to use text mining pipeline to postulate new relevant DDAs | Limited disease ontologies hinder the comparison and integration of data from different sources | Retrieve 5,443 unique diseases | Retrieve 12,263 co-mentions | Can retrieve 92% of all associations of a DDAs benchmark and still contribute with 16% of new DDAs |