Assessing surgical site infection risk factors using electronic medical records and text mining

Application of text mining in the biomedical domain

“…Michelson et al mined EMRs to detect surgical site infections (SSI) in unstructured clinical notes to improve SSI detection. 22 SSIs detected by traditional hospital-based surveillance were found using TM, along with an additional 37 SSIs not detected by traditional surveillance [122].…”

Identifying catheter-related events through sentence classification

“…Another interesting finding was that incomplete or inaccurate clinical notes hampered all methods, including manual chart review. In a 2014 study by Michelson et al (2014), text mining methods were found to be very effective in detecting different types of surgical site infections (SSIs). They did not consider CRBSIs specifically but their system was able to identify 100 % of infection cases detected by regular surveillance as well as 37 cases not previously identified.…”

Automated classification of BI-RADS in textual mammography reports

“…It is hypothesized that using TM to analyze textual records from the DTC and Picture archiving and communication system (PACS) may provide clinically reliable data [22][23][24] to predict BI-RADS levels. TM approaches have been used in medical research findings with variant targets including automatic disease classification of clinical discharges [25], recognition of patients' obesity case [26,27] and analysis of clinical documents to identify drug-disease associations [28][29][30][31][32].TM covers the gap between structured form of information and free-text [23,24] and uses Natural Language Processing (NLP) techniques, machine learning, and knowledge management to process free-text documents. On the other hand, TM has been used to transform essential data from text to logical and numerical format so they can be exported to data storage and analyzed [33].…”