The overarching long-term vision of our research is to create novel technologies for processing clinical free text. Such technologies will enable sophisticated and efficient indexing, retrieval and data mining over the ever increasing amounts of electronic clinical data. Processing free text poses a number of challenges to which the fields of Artificial intelligence, natural language processing and computer science in general have made advances. Methods for processing free text are informed by linguistic theory combined with the power of statistical inferencing. A key component to the next step, natural language understanding, is discovering events and their relations on a timeline. Temporal relations are of prime importance in biomedicine as they are intrinsically linked to diseases, signs and symptoms, and treatments. Understanding the timeline of clinically relevant events is key to the next generation of translational research where the importance of generalizing over large amounts of data holds the promise of deciphering biomedical puzzles. The goal of our current proposal is to discover temporal relations from clinical free text through achieving four specific aims:
Specific Aim 1 : Develop (1) a temporal relation annotation schema and guidelines for clinical free text based on TimeML, which will require extensions to Treebank, PropBank and VerbNet annotation guidelines to the clinical domain, (2) an annotated corpus following the temporal relations schema with additions to Treebank, PropBank and VerbNet, (3) a descriptive study comparing temporal relations in the clinical and general domains.
Specific Aim 2 : Extend and evaluate existing methods and/or develop new algorithms for temporal relation discovery in the clinical domain. Component-level evaluation Specific Aim 3: Integrate best method and/or a variety of methods for temporal relation discovery into the open source Mayo Clinic IE pipeline and release as open source annotators in the pipeline. Functional testing. Dissemination activities.
Specific Aim 4 : System-level evaluation. Test the functionality of the enhanced Mayo Clinic IE pipeline on translational research use cases, e.g. the progression of colon cancer as documented in clinical notes and pathology reports, the progression of brain tumor as documented in radiology reports. The methods we will use for the temporal relation discovery are based on machine learning, e.g., Support Vector Machine technology. Such methods require the annotation of a reference standard from which the computations are derived. The best methods will be released as part of the Mayo Clinic Information Extraction System for the larger community to use and contribute to. We will test the methods against biomedical queries.
(max 2-3 sentences) Temporal relations are of prime importance in biomedicine as they are intrinsically linked to diseases, signs and symptoms, and treatments. Understanding the timeline of clinically relevant events is key to the next generation of translational research where the importance of generalizing over large amounts of data holds the promise of deciphering biomedical puzzles. The goal of our current proposal is to automatically discover temporal relations from clinical free text and create a timeline.
|Lin, Chen; Dligach, Dmitriy; Miller, Timothy A et al. (2016) Multilayered temporal modeling for the clinical domain. J Am Med Inform Assoc 23:387-95|
|Pradhan, Sameer; Elhadad, NoÃ©mie; South, Brett R et al. (2015) Evaluating the state of the art in disorder recognition and normalization of the clinical narrative. J Am Med Inform Assoc 22:143-54|
|Lin, Chen; Karlson, Elizabeth W; Dligach, Dmitriy et al. (2015) Automatic identification of methotrexate-induced liver toxicity in patients with rheumatoid arthritis from the electronic medical record. J Am Med Inform Assoc 22:e151-61|
|Dligach, Dmitriy; Bethard, Steven; Becker, Lee et al. (2014) Discovering body site and severity modifiers in clinical texts. J Am Med Inform Assoc 21:448-54|
|Pfiffner, Pascal B; Oh, JiWon; Miller, Timothy A et al. (2014) ClinicalTrials.gov as a data source for semi-automated point-of-care trial eligibility screening. PLoS One 9:e111055|
|Albright, Daniel; Lanfranchi, Arrick; Fredriksen, Anwen et al. (2013) Towards comprehensive syntactic and semantic annotations of the clinical narrative. J Am Med Inform Assoc 20:922-30|