A natural language processing algorithm to extract characteristics of subdural hematoma from head CT reports
- 33 Downloads
Subdural hematoma (SDH) is the most common form of traumatic intracranial hemorrhage, and radiographic characteristics of SDH are predictive of complications and patient outcomes. We created a natural language processing (NLP) algorithm to extract structured data from cranial computed tomography (CT) scan reports for patients with SDH.
CT scan reports from patients with SDH were collected from a single center. All reports were based on cranial CT scan interpretations by board-certified attending radiologists. Reports were then coded by a pair of physicians for four variables: number of SDH, size of midline shift, thickness of largest SDH, and side of largest SDH. Inter-rater reliability was assessed. The annotated reports were divided into training (80%) and test (20%) datasets. Relevant information was extracted from text using a pattern-matching approach, due to the lack of a mention-level gold-standard corpus. Then, the NLP pipeline components were integrated using the Apache Unstructured Information Management Architecture. Output performance was measured as algorithm accuracy compared to the data coded by the two ED physicians.
A total of 643 scans were extracted. The NLP algorithm accuracy was high: 0.84 for side of largest SDH, 0.88 for thickness of largest SDH, and 0.92 for size of midline shift.
A NLP algorithm can structure key data from non-contrast head CT reports with high accuracy. The NLP is a potential tool to detect important radiographic findings from electronic health records, and, potentially, add decision support capabilities.
KeywordsSubdural hematoma Natural language processing Cranial CT reports Intracranial hemorrhage
Dr. Pruitt was supported by a National Research Service Award postdoctoral fellow supported by the Agency for Healthcare Research and Quality (AHRQ) T-32 HS 000078 (PI: Jane L. Holl, MD, MPH). AHRQ was not involved in the design or execution of this research. Dr. Pruitt is now supported by a career development award from the Society for Academic Emergency Medicine Foundation.
PP, AN, and WKT conceived of the study and designed the analysis. PB, JO, and PP participated in the abstraction and coding of data. WKT programmed the algorithm. PP performed the data analysis. PP drafted the manuscript and all authors contributed substantially to its revision. PP takes responsibility for the paper as a whole.
Compliance with ethical standards
Conflicts of interest
The authors declare that they have no conflict of interest.
- 4.Jain NL, Friedman C (1997) Identification of findings suspicious for breast cancer based on natural language processing of mammogram reports. Proc AMIA Annu Fall Symp: 829–833Google Scholar
- 5.Kreimeyer K, Foster M, Pandey A, Arya N, Halford G, Jones SF, Forshee R, Walderhaug M, Botsis T (2017) Natural language processing systems for capturing and standardizing unstructured clinical information: a systematic review. J Biomed Inform 73:14–29. https://doi.org/10.1016/j.jbi.2017.07.012 CrossRefGoogle Scholar
- 8.Kuo T-T, Rao P, Maehara C, et al (2016) Ensembles of NLP tools for data element extraction from clinical notes. AMIA. Annu Symp proceedings AMIA Symp 2016:1880–1889Google Scholar
- 9.Orlando A, Levy AS, Rubin BA, Tanner A, Carrick MM, Lieser M, Hamilton D, Mains CW, Bar-Or D (2018) Isolated subdural hematomas in mild traumatic brain injury. Part 2: a preliminary clinical decision support tool for neurosurgical intervention. J Neurosurg:1–8. https://doi.org/10.3171/2018.1.JNS171906
- 14.Savova GK, Masanz JJ, Ogren PV, Zheng J, Sohn S, Kipper-Schuler KC, Chute CG (2010) Mayo clinical text analysis and knowledge extraction system (cTAKES): architecture, component evaluation and applications. J Am Med Inform Assoc 17:507–513. https://doi.org/10.1136/jamia.2009.001560 CrossRefGoogle Scholar