Moonstone: a novel natural language processing system for inferring social risk from clinical narratives

Abstract

Background Social risk factors are important dimensions of health and are linked to access to care, quality of life, health outcomes and life expectancy. However, in the Electronic Health Record, data related to many social risk factors are primarily recorded in free-text clinical notes, rather than as more readily computable structured data, and hence cannot currently be easily incorporated into automated assessments of health. In this paper, we present Moonstone , a new, highly configurable rule-based clinical natural language processing system designed to automatically extract information that requires inferencing from clinical notes. Our initial use case for the tool is focused on the automatic extraction of social risk factor information — in this case, housing situation , living alone , and social support — from clinical notes. Nursing notes, social work notes, emergency room physician notes, primary care notes, hospital admission notes, and discharge summaries, all derived from the Veterans Health Administration, were used for algorithm development and evaluation. Results An evaluation of Moonstone demonstrated that the system is highly accurate in extracting and classifying the three variables of interest ( housing situation , living alone , and social support ). The system achieved positive predictive value (i.e. precision) scores ranging from 0.66 ( homeless/marginally housed ) to 0.98 ( lives at home/not homeless ), accuracy scores ranging from 0.63 ( lives in facility ) to 0.95 ( lives alone ), and sensitivity (i.e. recall) scores ranging from 0.75 ( lives in facility ) to 0.97 ( lives alone ). Conclusions The Moonstone system is — to the best of our knowledge — the first freely available, open source natural language processing system designed to extract social risk factors from clinical text with good ( lives in facility ) to excellent ( lives alone ) performance. Although developed with the social risk factor identification task in mind, Moonstone provides a powerful tool to address a range of clinical natural language processing tasks, especially those tasks that require nuanced linguistic processing in conjunction with inference capabilities.

Automated Coding of Under-Studied Medical Concept Domains: Linking Physical Activity Reports to the International Classification of Functioning, Disability, and Health

“…Under-studied domains of health information, such as social determinants of health or environmental exposures, generally lack welldeveloped vocabularies and terminologies that could otherwise guide extraction and coding of information from text. For example, a recent method for extracting social risk factors from narratives by Conway et al (66) utilizes handengineered word patterns to identify three types of risk factors, due to the lack of coverage of relevant terms in standardized vocabularies.…”

Meaningful Information in the Age of Big Data: A Scoping Review on Social Determinants of Health Data Collection for Electronic Health Records

“…(2) stress, or the effects of continuous anxiety, insecurity, and low selfesteem; (3) early life, or the impact that previous emotional and developmental experiences of childhood and adolescence have on adult health; (4) social exclusion, which is the effects of being treated less than equal as a result of discrimination, debilitation, racism, or stigmatization (such as is experienced by ex-convicts, the homeless, and those who are mentally ill); (5) work, or the specific contribution of stress at work to overall health; (6) unemployment, or the increased risk of premature death experienced by the unemployed and their families; (7) social support, including the impact that both emotional and tangible support systems have on health; (8) addiction, which encompasses the effects of alcohol, nicotine, and drug dependence both as a result of social inequality as well as a means of increasing its impact; (9) food, or how access to healthy foods can influence chronic disease management and progression; and (10) transport, encompassing both the ability to arrive at appointments and walk/exercise in safe environments. Residence, Living Situation, and Living Conditions Flowsheet prompts: "home", "house", "housing", "residence", "live", "living", "lives", "people", "mold", "insect", "rodent", "water", "heat", "social", "density" "Stairs", "railings", "safety", "safe", "facility", "group home", "skilled nursing facility", "assisted living facility", "support system", "family", "support", "housing conditions", "caregiver", "bathroom", "community support", "rehab", "assistive device", "social/environment", "equipment", "social support", "household", "transitional care", "social connectedness", "live alone" Flow Diagram…”

Measuring the Value of a Practical Text Mining Approach to Identify Patients With Housing Issues in the Free-Text Notes in Electronic Health Record: Findings of a Retrospective Cohort Study

“…Our results were slightly different from other studies using rule-based systems to identify social needs in free-text provider notes. For instance, Conway et al (55) tested the performance of Moonstone, a new, highly configurable rulebased clinical NLP system for extraction of information requiring inferencing from clinical notes derived from the Veterans Health Administration. Their system achieved a precision (positive predictive value) of 0.66 (lower than ∼94-96% at the phrase, note, and patient-level in our study) and a recall (sensitivity) of 0.87 (higher than ∼30-41% at phrase, note, and patient-level in our study) for phrases related to homelessness and marginally housed.…”