Limits...
Information discovery on electronic health records using authority flow techniques.

Hristidis V, Varadarajan RR, Biondich P, Weiner M - BMC Med Inform Decis Mak (2010)

Bottom Line: As the use of electronic health records (EHRs) becomes more widespread, so does the need to search and provide effective information discovery within them.Querying by keyword has emerged as one of the most effective paradigms for searching.We compare the effectiveness of two fundamentally different techniques for keyword search of EHRs.

View Article: PubMed Central - HTML - PubMed

Affiliation: School of Computing and Information Sciences, Florida International University, Miami, Florida, USA. vagelis@cis.fiu.edu

ABSTRACT

Background: As the use of electronic health records (EHRs) becomes more widespread, so does the need to search and provide effective information discovery within them. Querying by keyword has emerged as one of the most effective paradigms for searching. Most work in this area is based on traditional Information Retrieval (IR) techniques, where each document is compared individually against the query. We compare the effectiveness of two fundamentally different techniques for keyword search of EHRs.

Methods: We built two ranking systems. The traditional BM25 system exploits the EHRs' content without regard to association among entities within. The Clinical ObjectRank (CO) system exploits the entities' associations in EHRs using an authority-flow algorithm to discover the most relevant entities. BM25 and CO were deployed on an EHR dataset of the cardiovascular division of Miami Children's Hospital. Using sequences of keywords as queries, sensitivity and specificity were measured by two physicians for a set of 11 queries related to congenital cardiac disease.

Results: Our pilot evaluation showed that CO outperforms BM25 in terms of sensitivity (65% vs. 38%) by 71% on average, while maintaining the specificity (64% vs. 61%). The evaluation was done by two physicians.

Conclusions: Authority-flow techniques can greatly improve the detection of relevant information in EHRs and hence deserve further study.

Show MeSH

Related in: MedlinePlus

Sample User Survey Page for query "respiratory distress". Figure 4 shows a sample user survey page for query "respiratory distress". It starts with a brief explanation about the survey and then query results are displayed. Each result displays a brief description of the clinical entity in tabular format along with links to the complete description and adjacent entities.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2984470&req=5

Figure 4: Sample User Survey Page for query "respiratory distress". Figure 4 shows a sample user survey page for query "respiratory distress". It starts with a brief explanation about the survey and then query results are displayed. Each result displays a brief description of the clinical entity in tabular format along with links to the complete description and adjacent entities.

Mentions: To gain preliminary insights about CO vs. BM25, we worked with a convenience sample of two primary-care physicians (MW and PB), who were blinded to the automated rankings. Note that the users have been trained before since we also conducted an initial version of the User Survey for the users to get used to the way the survey was built. In addition to this, we also allowed users to play with a demonstration website that permitted them the key in any set of keywords and see the full description of results along with the corresponding explaining sub graph of the result. For each keyword query, the top 5 results from each algorithm (BM25, CO085, CO030, and CO085BM25) were computed and merged to a single list of results for that query. The information about results and the corresponding algorithms was hidden from the users. Then, for each keyword query, the user was asked to select the top 5 results that were both relevant and important for each query. To help the users evaluate the results, next to each result we displayed the links "Full Description" and "Adjacent Entities", to show the details of the entity, as well as the relationship of this entity to query keywords and the rest of the clinical web, respectively. Figure 4 shows a sample user survey page for query "respiratory distress". Figure 5 shows a sample "Full Description" page of a hospitalization object for query "respiratory distress".


Information discovery on electronic health records using authority flow techniques.

Hristidis V, Varadarajan RR, Biondich P, Weiner M - BMC Med Inform Decis Mak (2010)

Sample User Survey Page for query "respiratory distress". Figure 4 shows a sample user survey page for query "respiratory distress". It starts with a brief explanation about the survey and then query results are displayed. Each result displays a brief description of the clinical entity in tabular format along with links to the complete description and adjacent entities.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2984470&req=5

Figure 4: Sample User Survey Page for query "respiratory distress". Figure 4 shows a sample user survey page for query "respiratory distress". It starts with a brief explanation about the survey and then query results are displayed. Each result displays a brief description of the clinical entity in tabular format along with links to the complete description and adjacent entities.
Mentions: To gain preliminary insights about CO vs. BM25, we worked with a convenience sample of two primary-care physicians (MW and PB), who were blinded to the automated rankings. Note that the users have been trained before since we also conducted an initial version of the User Survey for the users to get used to the way the survey was built. In addition to this, we also allowed users to play with a demonstration website that permitted them the key in any set of keywords and see the full description of results along with the corresponding explaining sub graph of the result. For each keyword query, the top 5 results from each algorithm (BM25, CO085, CO030, and CO085BM25) were computed and merged to a single list of results for that query. The information about results and the corresponding algorithms was hidden from the users. Then, for each keyword query, the user was asked to select the top 5 results that were both relevant and important for each query. To help the users evaluate the results, next to each result we displayed the links "Full Description" and "Adjacent Entities", to show the details of the entity, as well as the relationship of this entity to query keywords and the rest of the clinical web, respectively. Figure 4 shows a sample user survey page for query "respiratory distress". Figure 5 shows a sample "Full Description" page of a hospitalization object for query "respiratory distress".

Bottom Line: As the use of electronic health records (EHRs) becomes more widespread, so does the need to search and provide effective information discovery within them.Querying by keyword has emerged as one of the most effective paradigms for searching.We compare the effectiveness of two fundamentally different techniques for keyword search of EHRs.

View Article: PubMed Central - HTML - PubMed

Affiliation: School of Computing and Information Sciences, Florida International University, Miami, Florida, USA. vagelis@cis.fiu.edu

ABSTRACT

Background: As the use of electronic health records (EHRs) becomes more widespread, so does the need to search and provide effective information discovery within them. Querying by keyword has emerged as one of the most effective paradigms for searching. Most work in this area is based on traditional Information Retrieval (IR) techniques, where each document is compared individually against the query. We compare the effectiveness of two fundamentally different techniques for keyword search of EHRs.

Methods: We built two ranking systems. The traditional BM25 system exploits the EHRs' content without regard to association among entities within. The Clinical ObjectRank (CO) system exploits the entities' associations in EHRs using an authority-flow algorithm to discover the most relevant entities. BM25 and CO were deployed on an EHR dataset of the cardiovascular division of Miami Children's Hospital. Using sequences of keywords as queries, sensitivity and specificity were measured by two physicians for a set of 11 queries related to congenital cardiac disease.

Results: Our pilot evaluation showed that CO outperforms BM25 in terms of sensitivity (65% vs. 38%) by 71% on average, while maintaining the specificity (64% vs. 61%). The evaluation was done by two physicians.

Conclusions: Authority-flow techniques can greatly improve the detection of relevant information in EHRs and hence deserve further study.

Show MeSH
Related in: MedlinePlus