Transcript Document
Automated Vocabulary Maintenance System for the Open Access, Collaborative Consumer Health Vocabulary Kristina M Doing-Harris, BCompSci, MA, MS, PhD; Qing Zeng-Treitler, PhD Department of Biomedical Informatics, University of Utah, Salt Lake City, USA Introduction •Controlled vocabularies play an important role in the development of biomedical informatics applications. •Consumer health vocabulary (CHV), has been rising in prominence. •Controlled vocabularies require maintenance and update, due to the continuing evolution of language itself. •In healthcare especially there is a constant stream of new names (e.g. new medications, disorders, tests) being coined in the literature. • CHV must keep up with these changes in the language used by consumers. Main Question How can a consumer health vocabulary evolve with consumer language? Schematic Diagram of the AVM system PatientsLikeMe : Patients Helping Patients Live Better Every Day. Secure login Join today! You appear to have JavaScript disabled in your browser. PatientsLikeMe relies on JavaScript and Cookies to deliver the best possible experience to you. How do I enable JavaScript? Find Patients Just Like You I wish this site was around years ago as I lost so much time and money doing what didn't work. Multiple Sclerosis Community Member ; Find a patient like you now Current Disease Communities Prevalent Diseases ALS/MND ID 51 52 53 54 55 56 Stage 1 (A,B & C) 57 58 59 PatientsLikeMe.com Raw text file excerpt ID Term 60 firstPOS_N 1 1 1 … 1… 1… 1… 0 0 0 1 0 1 Frequency … … … 2560 … 490 … 228 … 34 34 1 0 1 0 1 0 … 34 … 0 … … … … 1… freq_in_subs 0 1214 83 9428 9394 9546 … 34 34 34 … … … 42 … 9360 9478 9614 27 Excerpt from n-gram database CUI 68 Eszopiclone 164 Piroxicam 214 Back pain 366 Adherence 402 Celecoxib 403 diagnose medical conditions 404 using Ankle-Foot Orthosis 405 60 mg Type ALS Motor Neuron Disease 406 tom Sex 407 Nebulizer Treatment device 408 Efficacy Reasons taken of patients Aquatic Therapy Exercises 409 Treatment Report 410 See all patient evaluation Ngram in_NP initfunction pageInit100 shortlinks35 over navigation PatientsLikeMe Share your real-world symptom navigation PatientsLikeMe Share your real-world symptom navigation PatientsLikeMe Share your real-world symptom experience PatientsLikeMe Share your realworld symptom PatientsLikeMe Share your realworld symptom experience PatientsLikeMe Share your realworld symptom experience with Share your real-world symptom isTerm C1436328 C0031990 C0004604 cScore Frequency inGoldStand inMedRec 0 0 0 3.6203 3.6203 1.894 3.6203 3.6203 3.3945 3.366 3.2662 0 0 9 0 0 0 0 31 31 61 9 1032 31 10 9 31 1 0 1 1 1 0 0 0 1 0 0 1 1 0 0 0 0 3.3945 0 9 0 0 0 0 3.6203 9.509775004 2.3304 0 9 131 1 0 0 0 0 4.0203 11.60964047 5 0 0 0 3.3945 20 0 0 0 C0538927 0 Excerpt from potential term database Stage 2 (C) CHV Update Wiki www.ConsumerHealthVocab.utah.edu/AutoVocabMaint Results •Combined: Termhood score threshold of 3.6 for terms found in the medical records and C-value threshold of 15. •Produced 774 candidate terms, with 237 valid terms. •Reviewers will find 1 valid term for every 3 or 4 candidate terms. •Better than initial n-gram list with an average of 1 valid term for every 137 candidate terms. Summary of Conclusions • Social network data can be used to provide a living corpus. • It can be mined to provide new consumer health vocabulary terms. • Using ATR and dictionary look up can produce a concise list of candidate terms. • Allowing the consumer health vocabulary to evolve with consumer language. CHV Website www.ConsumerHealthVocab.org Acknowledgements NLM Training Grant No. RO1 LM07222 Contact Information [email protected]