Constraints Driven Information Extraction in the Medical Domain Dan Roth Department of Computer Science University of Illinois at Urbana-Champaign November 2012 With thanks to: Collaborators: Ming-Wei Chang, Prateek Jindal, Lev Ratinov, Many others Funding: NSF; DHS; NIH; DARPA.ONC HITC Workshop 2012 @ University of Illinoi Page 1
Technological Challenges Medical Informatics Privacy Challenges An electronic health record (EHR) is a personal health record in digital format. Patient-centric information that should aid clinical decision-making. Includes information relating to the current and historical health, medical conditions and medical tests of its subject. Data about medical referrals, treatments, medications, demographic information and other non-clinical administrative information. A narrative with embedded database elements Potential Benefits Health Needs Utilize in medical advice systems Enable information extraction Medication selection and tracking (Vioxx…) & information integration Disease outbreak and control across various projections Science of the data and across Correlating response to drugs with other systems conditions 2
Analyzing Electronic Health Records Identify Important Mentions [The patient] is a 65 year old female with [post thoracotomy syndrome] [that] The patient is a 65 year old female with post thoracotomy syndrome that occurred on the site of [[her] thoracotomy incision] . occurred on the site of her thoracotomy incision . [She] had [a thoracic aortic aneurysm] repaired in the past and subsequently She had a thoracic aortic aneurysm repaired in the past and subsequently developed [neuropathic pain] at [the incision site] . developed neuropathic pain at the incision site . [She] is currently on [Vicodin] , one to two tablets every four hours p.r.n. , She is currently on Vicodin , one to two tablets every four hours p.r.n. , [Fentanyl patch] 25 mcg an hour , change of patch every 72 hours , [Elavil] 50 Fentanyl patch 25 mcg an hour , change of patch every 72 hours , Elavil 50 mgq .h.s. , [Neurontin] 600 mg p.o. t.i.d. with still what [she] reports as mgq .h.s. , Neurontin 600 mg p.o. t.i.d. with still what she reports as stabbing [stabbing left-sided chest pain] [that] can be as severe as a 7/10. left-sided chest pain that can be as severe as a 7/10. [She] has failed [conservative therapy] and is admitted for [a spinal cord She has failed conservative therapy and is admitted for a spinal cord stimulator trial] . stimulator trial . 3
Red : Problems Analyzing Electronic Health Records Green : Treatments Purple : Tests Identify Concept Types Blue : People [The patient] is a 65 year old female with [post thoracotomy syndrome] [that] occurred on the site of [[her] thoracotomy incision] . [She] had [a thoracic aortic aneurysm] repaired in the past and subsequently developed [neuropathic pain] at [the incision site] . [She] is currently on [Vicodin] , one to two tablets every four hours p.r.n. , [Fentanyl patch] 25 mcg an hour , change of patch every 72 hours , [Elavil] 50 mgq .h.s. , [Neurontin] 600 mg p.o. t.i.d. with still what [she] reports as [stabbing left-sided chest pain] [that] can be as severe as a 7/10. [She] has failed [conservative therapy] and is admitted for [a spinal cord stimulator trial] . 4
Other needs: temporal Analyzing Electronic Health Records recognition & reasoning, relations, quantities, etc. Coreference Resolution [The patient] is a 65 year old female with [post thoracotomy syndrome] [that] occurred on the site of [[her] thoracotomy incision] . [She] had [a thoracic aortic aneurysm] repaired in the past and subsequently developed [neuropathic pain] at [the incision site] . [She] is currently on [Vicodin] , one to two tablets every four hours p.r.n. , [Fentanyl patch] 25 mcg an hour , change of patch every 72 hours , [Elavil] 50 mgq .h.s. , [Neurontin] 600 mg p.o. t.i.d. with still what [she] reports as [stabbing left-sided chest pain] [that] can be as severe as a 7/10. [She] has failed [conservative therapy] and is admitted for [a spinal cord stimulator trial] . 5
Multiple Applications Clinical Decisions: “Please show me the reports of all patients who had headache that was not cured by Aspirin.” Concept Recognition; Relation Identification (Problem, Treatment) “Please show me the reports of all patients who have had myocardial infarction (heart attack) more than once.” Coreference Resolution Identification of sensitive data (Privacy Reasons) HIV Data, Drug Abuse, Family Abuse, Genetic Information Concept Recognition, Relations Recognition (drug, drug abuse), coreference resolution (multiple incidents, same people) Generating summaries for patient Creating automatic reminders of medications 6
Machine Learning + Inference based NLP All these Information extraction problems are being studied in Natural Language Processing, typically on newswire data These problems are extremely difficult due to Ambiguity (everything has multiple meanings) Variability (everything you want to say you can say in many ways) Impossible to reliably “program” predicates of interest. Models are based on Statistical Machine Learning & Inference The focus is on modeling and learning algorithms for different phenomena Classification models Well understood; easy to build black box categorizers Structured models Learning protocols that exploit Minimal Supervision Inference as a way to introduce domain & task specific constraints 7
Information Extraction in the Medical Domain Models learned on newswire data do not adapt well to the medical domain. Different vocabulary, sentence and document structure. More importantly, the medical domain offers a chance to do better than the general newswire domain. Background Knowledge: Narrow domain; a lot of manually curated KB resources that can be used to help identification & disambiguation. UMLS: A large biomedical KB, with semantic types and relationships between concepts. Mesh: A large thesaurus of medical vocabulary. SNOMED CT: A comprehensive clinical terminology. Structure: Medical Text has more structure that can be exploited. Discourse structure: Concepts in the section “Principal Diagnosis” are more likely to be “medical problems”. HERs have some internal structure: Doctors, One Patient, Family Members. 8
Learning and Inference for Medical Information Extraction We develop models that make global decisions that consist of several local decisions — on identified concepts, relations between identified concepts, etc. Our models are developed in such a way that they can exploit external Knowledge that could relate sub-problems Allows local models to retract/modify decisions that do not cohere with decisions made by other local models. Goal: Incorporate local models’ information, along with prior knowledge (constraints) in making coherent decisions Decisions that respect the local models as well as domain & context specific knowledge/constraints. Page 9
Constrained Conditional Models Penalty for violating the constraint. (Soft) constraints component Weight Vector for “local” models How far y is from Features, classifiers; log- a “legal” assignment linear models (HMM, CRF) or a combination A collection of local models Knowledge as Constraints Coreference: pairwise classifier Doctor cannot co-ref with a patient. between mentions Consistency with KB resources Concepts: a model that determines Consistency of anatomical terms boundaries for important phrases. Legitimacy of relations Relations: Per-relation classifier Page 10
Constrained Conditional Models Penalty for violating the constraint. (Soft) constraints component Weight Vector for “local” models How far y is from Features, classifiers; log- a “legal” assignment linear models (HMM, CRF) or a combination How to solve? How to train? This is an Integer Linear Program Training is learning the objective function Solving using ILP packages gives an exact solution. We decompose the objective and learn components. (Other methods are possible) search techniques are possible ½ i s can be learned jointly 1: 11
Current Status State-of-the-art Coreference Resolution System for Clinical Narratives (JAMIA’12, COLING’12) State-of-the-art Concept and Relation Extraction (I2B2 workshop’12) Current work: Continuing work on concept identification and Relations End-2-End Coreference Resolution System Sensitive Concepts 12
Thanks! Mapping to Encyclopedic Resources (Demo) Beyond supporting better Natural Language Processing, Wikification could allow people to read and understand these documents and access them in an easier way. http://en.wikipedia.org/wiki/Amitriptyline Hydrocodone/paracetamol http://http://en.wikipedia.org/wiki/Vicodin 13
Recommend
More recommend
