@article{RN410,
   author = {Seinen, T. M. and Kors, J. A. and van Mulligen, E. M. and Rijnbeek, P. R.},
   title = {Using Structured Codes and Free-Text Notes to Measure Information Complementarity in Electronic Health Records: Feasibility and Validation Study},
   journal = {J Med Internet Res},
   volume = {27},
   pages = {e66910},
   note = {1438-8871 Seinen, Tom M Orcid: 0000-0001-5369-8260 Kors, Jan A Orcid: 0000-0002-4929-026x van Mulligen, Erik M Orcid: 0000-0003-1377-9386 Rijnbeek, Peter R Orcid: 0000-0003-0621-1979 Journal Article Validation Study Canada 2025/02/13 23:03 J Med Internet Res. 2025 Feb 13;27:e66910. doi: 10.2196/66910.},
   abstract = {BACKGROUND: Electronic health records (EHRs) consist of both structured data (eg, diagnostic codes) and unstructured data (eg, clinical notes). It is commonly believed that unstructured clinical narratives provide more comprehensive information. However, this assumption lacks large-scale validation and direct validation methods. OBJECTIVE: This study aims to quantitatively compare the information in structured and unstructured EHR data and directly validate whether unstructured data offers more extensive information across a patient population. METHODS: We analyzed both structured and unstructured data from patient records and visits in a large Dutch primary care EHR database between January 2021 and January 2024. Clinical concepts were identified from free-text notes using an extraction framework tailored for Dutch and compared with concepts from structured data. Concept embeddings were generated to measure semantic similarity between structured and extracted concepts through cosine similarity. A similarity threshold was systematically determined via annotated matches and minimized weighted Gini impurity. We then quantified the concept overlap between structured and unstructured data across various concept domains and patient populations. RESULTS: In a population of 1.8 million patients, only 13% of extracted concepts from patient records and 7% from individual visits had similar structured counterparts. Conversely, 42% of structured concepts in records and 25% in visits had similar matches in unstructured data. Condition concepts had the highest overlap, followed by measurements and drug concepts. Subpopulation visits, such as those with chronic conditions or psychological disorders, showed different proportions of data overlap, indicating varied reliance on structured versus unstructured data across clinical contexts. CONCLUSIONS: Our study demonstrates the feasibility of quantifying the information difference between structured and unstructured data, showing that the unstructured data provides important additional information in the studied database and populations. The annotated concept matches are made publicly available for the clinical natural language processing community. Despite some limitations, our proposed methodology proves versatile, and its application can lead to more robust and insightful observational clinical research.},
   keywords = {*Electronic Health Records
Humans
Feasibility Studies
Netherlands
Ehr
clinical concept extraction
clinical concept similarity
code
data
electronic health records
electronic record
framework
free-text
information
machine learning
named entity recognition
natural language processing
patient records
structured data
text mining
unstructured data
word embeddings ,
ISSN = 1439-4456 (Print)
1438-8871 ,
DOI = 10.2196/66910},
   year = {2025},
   type = {Journal Article}
}

{"_id":"srd3Kv8eLrxMwGnYj","bibbaseid":"seinen-kors-vanmulligen-rijnbeek-usingstructuredcodesandfreetextnotestomeasureinformationcomplementarityinelectronichealthrecordsfeasibilityandvalidationstudy-2025","author_short":["Seinen, T. M.","Kors, J. A.","van Mulligen, E. M.","Rijnbeek, P. R."],"bibdata":{"bibtype":"article","type":"Journal Article","author":[{"propositions":[],"lastnames":["Seinen"],"firstnames":["T.","M."],"suffixes":[]},{"propositions":[],"lastnames":["Kors"],"firstnames":["J.","A."],"suffixes":[]},{"propositions":["van"],"lastnames":["Mulligen"],"firstnames":["E.","M."],"suffixes":[]},{"propositions":[],"lastnames":["Rijnbeek"],"firstnames":["P.","R."],"suffixes":[]}],"title":"Using Structured Codes and Free-Text Notes to Measure Information Complementarity in Electronic Health Records: Feasibility and Validation Study","journal":"J Med Internet Res","volume":"27","pages":"e66910","note":"1438-8871 Seinen, Tom M Orcid: 0000-0001-5369-8260 Kors, Jan A Orcid: 0000-0002-4929-026x van Mulligen, Erik M Orcid: 0000-0003-1377-9386 Rijnbeek, Peter R Orcid: 0000-0003-0621-1979 Journal Article Validation Study Canada 2025/02/13 23:03 J Med Internet Res. 2025 Feb 13;27:e66910. doi: 10.2196/66910.","abstract":"BACKGROUND: Electronic health records (EHRs) consist of both structured data (eg, diagnostic codes) and unstructured data (eg, clinical notes). It is commonly believed that unstructured clinical narratives provide more comprehensive information. However, this assumption lacks large-scale validation and direct validation methods. OBJECTIVE: This study aims to quantitatively compare the information in structured and unstructured EHR data and directly validate whether unstructured data offers more extensive information across a patient population. METHODS: We analyzed both structured and unstructured data from patient records and visits in a large Dutch primary care EHR database between January 2021 and January 2024. Clinical concepts were identified from free-text notes using an extraction framework tailored for Dutch and compared with concepts from structured data. Concept embeddings were generated to measure semantic similarity between structured and extracted concepts through cosine similarity. A similarity threshold was systematically determined via annotated matches and minimized weighted Gini impurity. We then quantified the concept overlap between structured and unstructured data across various concept domains and patient populations. RESULTS: In a population of 1.8 million patients, only 13% of extracted concepts from patient records and 7% from individual visits had similar structured counterparts. Conversely, 42% of structured concepts in records and 25% in visits had similar matches in unstructured data. Condition concepts had the highest overlap, followed by measurements and drug concepts. Subpopulation visits, such as those with chronic conditions or psychological disorders, showed different proportions of data overlap, indicating varied reliance on structured versus unstructured data across clinical contexts. CONCLUSIONS: Our study demonstrates the feasibility of quantifying the information difference between structured and unstructured data, showing that the unstructured data provides important additional information in the studied database and populations. The annotated concept matches are made publicly available for the clinical natural language processing community. Despite some limitations, our proposed methodology proves versatile, and its application can lead to more robust and insightful observational clinical research.","keywords":"*Electronic Health Records Humans Feasibility Studies Netherlands Ehr clinical concept extraction clinical concept similarity code data electronic health records electronic record framework free-text information machine learning named entity recognition natural language processing patient records structured data text mining unstructured data word embeddings , ISSN = 1439-4456 (Print) 1438-8871 , DOI = 10.2196/66910","year":"2025","bibtex":"@article{RN410,\r\n author = {Seinen, T. M. and Kors, J. A. and van Mulligen, E. M. and Rijnbeek, P. R.},\r\n title = {Using Structured Codes and Free-Text Notes to Measure Information Complementarity in Electronic Health Records: Feasibility and Validation Study},\r\n journal = {J Med Internet Res},\r\n volume = {27},\r\n pages = {e66910},\r\n note = {1438-8871 Seinen, Tom M Orcid: 0000-0001-5369-8260 Kors, Jan A Orcid: 0000-0002-4929-026x van Mulligen, Erik M Orcid: 0000-0003-1377-9386 Rijnbeek, Peter R Orcid: 0000-0003-0621-1979 Journal Article Validation Study Canada 2025/02/13 23:03 J Med Internet Res. 2025 Feb 13;27:e66910. doi: 10.2196/66910.},\r\n abstract = {BACKGROUND: Electronic health records (EHRs) consist of both structured data (eg, diagnostic codes) and unstructured data (eg, clinical notes). It is commonly believed that unstructured clinical narratives provide more comprehensive information. However, this assumption lacks large-scale validation and direct validation methods. OBJECTIVE: This study aims to quantitatively compare the information in structured and unstructured EHR data and directly validate whether unstructured data offers more extensive information across a patient population. METHODS: We analyzed both structured and unstructured data from patient records and visits in a large Dutch primary care EHR database between January 2021 and January 2024. Clinical concepts were identified from free-text notes using an extraction framework tailored for Dutch and compared with concepts from structured data. Concept embeddings were generated to measure semantic similarity between structured and extracted concepts through cosine similarity. A similarity threshold was systematically determined via annotated matches and minimized weighted Gini impurity. We then quantified the concept overlap between structured and unstructured data across various concept domains and patient populations. RESULTS: In a population of 1.8 million patients, only 13% of extracted concepts from patient records and 7% from individual visits had similar structured counterparts. Conversely, 42% of structured concepts in records and 25% in visits had similar matches in unstructured data. Condition concepts had the highest overlap, followed by measurements and drug concepts. Subpopulation visits, such as those with chronic conditions or psychological disorders, showed different proportions of data overlap, indicating varied reliance on structured versus unstructured data across clinical contexts. CONCLUSIONS: Our study demonstrates the feasibility of quantifying the information difference between structured and unstructured data, showing that the unstructured data provides important additional information in the studied database and populations. The annotated concept matches are made publicly available for the clinical natural language processing community. Despite some limitations, our proposed methodology proves versatile, and its application can lead to more robust and insightful observational clinical research.},\r\n keywords = {*Electronic Health Records\r\nHumans\r\nFeasibility Studies\r\nNetherlands\r\nEhr\r\nclinical concept extraction\r\nclinical concept similarity\r\ncode\r\ndata\r\nelectronic health records\r\nelectronic record\r\nframework\r\nfree-text\r\ninformation\r\nmachine learning\r\nnamed entity recognition\r\nnatural language processing\r\npatient records\r\nstructured data\r\ntext mining\r\nunstructured data\r\nword embeddings ,\r\nISSN = 1439-4456 (Print)\r\n1438-8871 ,\r\nDOI = 10.2196/66910},\r\n year = {2025},\r\n type = {Journal Article}\r\n}\r\n\r\n","author_short":["Seinen, T. M.","Kors, J. A.","van Mulligen, E. M.","Rijnbeek, P. R."],"key":"RN410","id":"RN410","bibbaseid":"seinen-kors-vanmulligen-rijnbeek-usingstructuredcodesandfreetextnotestomeasureinformationcomplementarityinelectronichealthrecordsfeasibilityandvalidationstudy-2025","role":"author","urls":{},"keyword":["*Electronic Health Records Humans Feasibility Studies Netherlands Ehr clinical concept extraction clinical concept similarity code data electronic health records electronic record framework free-text information machine learning named entity recognition natural language processing patient records structured data text mining unstructured data word embeddings","ISSN = 1439-4456 (Print) 1438-8871","DOI = 10.2196/66910"],"metadata":{"authorlinks":{}},"html":""},"bibtype":"article","biburl":"https://bibbase.org/network/files/qwipow2DJsAxqNp83","dataSources":["vDmJPYmHEKAT3tMNL","PcZMFHooKwpboi2he","jZMXEeHTfokmisJDv"],"keywords":["*electronic health records humans feasibility studies netherlands ehr clinical concept extraction clinical concept similarity code data electronic health records electronic record framework free-text information machine learning named entity recognition natural language processing patient records structured data text mining unstructured data word embeddings","issn = 1439-4456 (print) 1438-8871","doi = 10.2196/66910"],"search_terms":["using","structured","codes","free","text","notes","measure","information","complementarity","electronic","health","records","feasibility","validation","study","seinen","kors","van mulligen","rijnbeek"],"title":"Using Structured Codes and Free-Text Notes to Measure Information Complementarity in Electronic Health Records: Feasibility and Validation Study","year":2025}