BACKGROUND/RATIONALE:
Post-deployment homelessness has been a major issue for Veterans after all conflicts and has been a priority area for the VA. The exact number of homeless Veterans is unknown; the US Department of Veterans Affairs estimates that at least 131,000 Veterans are homeless every night in the US. There is an urgent need to develop electronic algorithms to identify homeless Veterans and identify those Veterans at risk for homelessness. These alerts will be triggered by mining structured and unstructured (free text) data elements for known risk factors in the electronic medical record using natural language processing methods. The program will focus on male and female OEF/OIF Veterans and is intended to complement and enhance current local and national VA initiatives to address homelessness among Veterans.

OBJECTIVE(S):
The objective of this project is to develop and validate algorithms using clinical narratives and structured data to flag Veterans who are homeless or at high risk of homelessness.

METHODS:
(1) Identify a cohort of Veterans whose homeless status has been established
(2) Develop a vocabulary to identify concepts, features and documentation related to homelessness in the VA electronic medical record with special emphasis on psychosocial phenotyping
(3) Develop electronic algorithms to identify Veterans who are homeless or at risk of homelessness, using a domain-specific lexicon and natural language processing methods
(4) Perform retrospective validation of the algorithms using national VA electronic data
(5) Establish working relationships with community homeless service providers in Salt Lake City, Utah.

FINDINGS/RESULTS:
1. Developing a lexicon: We generated a human-curated lexicon for concepts related to homelessness. This is an important contribution to the field as there was no readily available lexicon for this domain.

2. Information (concept) extraction using natural language processing (NLP): We have used this lexicon to develop an NLP algorithm that extracts concepts related to homelessness from VA electronic records. We have trained and tested the algorithm on a human-reviewed reference standard set of medical notes that contained these concepts. The overall performance (positive predictive value, PPV) of the algorithm on this reference standard set is 77%.

3. Information retrieval for Veteran homelessness: An off-the-shelf VA developed tool, Automated Retrieval Console (ARC) was successfully adapted to the homelessness domain. The ARC tool was trained to perform document level classification. Performance has been measured at a precision of 94.5, recall of 95.2, and F-measure of 94.8.

4. Early identification of concepts related to homelessness in free text: We tested the hypothesis that concepts related to homelessness written in the free text of the medical record would precede the identification of homelessness by administrative data (ICD-9-CM). We applied our natural language processing algorithms for detecting homelessness and risk factors to medical notes from 50 randomly selected Veterans who were found to be homeless using the standard VA administrative data case definition. Notes from a control group of 50 Veterans who did not have an administrative indicator for homelessness were also processed.
'Direct evidence' of homelessness appeared in the notes of 30% of homeless Veterans a month or more before an administrative code for homelessness. Notes from 88% had evidence of risk factors related to homelessness prior to receiving an administrative code for homelessness. Among the notes of non-homeless Veterans, only 1 had 'direct evidence' and 6 had 'indirect evidence'.

5. Scaling up of NLP concept extraction algorithms: We set out to develop algorithms to improve efficiency of patient phenotyping using NLP on large corpora of text data. We sought to determine the note titles in the database with highest yield and precision for psychosocial concepts. We used our lexicon for homelessness risk factors as a basis for this work.
From a database of over 1 billion documents from VA medical facilities, a random sample of 1500 documents from each of 218 enterprise note titles were chosen. Psychosocial concepts were extracted using a UIMA-AS based NLP pipeline, using a lexicon of relevant concepts with negation and template format annotators. Human reviewers evaluated a subset of documents for false positives. High yield documents were identified by hit rate and precision. Reasons for false positivity were characterized.
A total of 58,707 psychosocial concepts were identified from 316,355 documents for an overall hit rate of 0.2 concepts per document (median 0.1, range 1.6 to 0). Of 6031 concepts reviewed from a high yield set of note titles, the overall precision for all concept categories was 80% with variability among note titles and concept categories. Reasons for false positivity included templating, negation, context and alternate meaning of words.

6. Identification of patterns in resource utilization prior to administrative recognition of homelessness: There are limited data on resources utilized by Veterans prior to their identification as being homeless. We performed visual analytics on longitudinal medical encounter data prior to the official recognition of homelessness in a large cohort of OEF/OIF Veterans. A statistically significant increase in numbers of visits in the immediate 30 days prior to the recognition of homelessness was noted as compared to an earlier period. Further studies are ongoing to validate this novel finding as a predictive tool.

7. Hepatitis C and homeless Veterans:We sought to describe the rates and predictors of initiation of treatment for chronic hepatitis C virus infection (HCV) in a cohort of HCV positive US Veterans with evidence of homelessness. Rates of treatment among homeless and non-homeless HCV Veterans were very low and clinically similar, though statistically significant. (6.2% vs. 7.4%, p<0.0001). Patients age 50, those with drug abuse, diabetes and hemoglobin < 10 g/dL were less likely to be treated. Genotype 2/3 increased the likelihood of treatment.

8. Pneumonia and homeless individuals: We evaluated the admission decisions and outcomes in homeless individuals diagnosed with community acquired pneumonia (CAP) seen at an urban community hospital. A large cohort of homeless patients with CAP demonstrated higher hospitalization risk but similar lengths of stay and costs as nonhomeless patients.

9. We have established excellent working relationships with our local community homeless service providers.

IMPACT:
Detecting homelessness or identifying Veterans at risk for homelessness is an important target for sentinel event surveillance. The benefits of such surveillance are multifold. In addition, this work forms the foundation for a currently funded HSR&D grant whose goal is to develop automated predictive models to identify Veterans at risk of homelessness.

---

External Links for this Project

---

PUBLICATIONS:

Journal Articles

1. Gundlapalli AV, Nelson RE, Haroldsen C, Carter ME, LaFleur J. Correlates of Initiation of Treatment for Chronic Hepatitis C Infection in United States Veterans, 2004-2009. PLoS ONE. 2015 Jul 13; 10(7):e0132056. [view]
2. Redd A, Carter M, Divita G, Shen S, Palmer M, Samore M, Gundlapalli AV. Detecting earlier indicators of homelessness in the free text of medical records. Studies in health technology and informatics. 2014 Jan 1; 202:153-6. [view]
3. Gundlapalli AV, Divita G, Redd A, Carter ME, Ko D, Rubin M, Samore M, Strymish J, Krein S, Gupta K, Sales A, Trautner BW. Detecting the presence of an indwelling urinary catheter and urinary symptoms in hospitalized patients using natural language processing. Journal of Biomedical Informatics. 2017 Jul 1; 71S:S39-S45. [view]
4. Skelton F, Campbell B, Horwitz D, Krein S, Sales A, Gundlapalli A, Trautner BW. Developing a user-friendly report for electronically assisted surveillance of catheter-associated urinary tract infection. American journal of infection control. 2017 May 1; 45(5):572-574. [view]
5. Huttner B, Jones M, Rubin MA, Neuhauser MM, Gundlapalli A, Samore M. Drugs of last resort? The use of polymyxins and tigecycline at US Veterans Affairs medical centers, 2005-2010. PLoS ONE. 2013 Jun 24; 7(5):e36649. [view]
6. de Perio MA, Brueck SE, Mueller CA, Milne CK, Rubin MA, Gundlapalli AV, Mayer J. Evaluation of 2009 pandemic influenza A (H1N1) exposures and illness among physicians in training. American journal of infection control. 2012 Sep 1; 40(7):617-21. [view]
7. Tran LT, Divita G, Carter ME, Judd J, Samore MH, Gundlapalli AV. Exploiting the UMLS Metathesaurus for extracting and categorizing concepts representing signs and symptoms to anatomically related organ systems. Journal of Biomedical Informatics. 2015 Dec 1; 58:19-27. [view]
8. Gundlapalli AV, Redd A, Carter ME, Palmer M, Peterson R, Samore MH. Exploring patterns in resource utilization prior to the formal identification of homelessness in recently returned veterans. Studies in health technology and informatics. 2014 Jan 1; 202:265-8. [view]
9. Gundlapalli AV, Carter ME, Divita G, Shen S, Palmer M, South B, Durgahee BS, Redd A, Samore M. Extracting Concepts Related to Homelessness from the Free Text of VA Electronic Medical Records. AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2014 Nov 14; 2014:589-98. [view]
10. Carter ME, Divita G, Redd A, Rubin MA, Samore MH, Gupta K, Trautner BW, Gundlapalli AV. Finding 'Evidence of Absence' in Medical Notes: Using NLP for Clinical Inferencing. Studies in health technology and informatics. 2016 Jan 1; 226:79-82. [view]
11. Gundlapalli AV, Redd D, Gibson BS, Carter M, Korhonen C, Nebeker J, Samore MH, Zeng-Treitler Q. Maximizing clinical cohort size using free text queries. Computers in biology and medicine. 2015 May 1; 60:1-7. [view]
12. Gundlapalli AV, Fargo JD, Metraux S, Carter ME, Samore MH, Kane V, Culhane DP. Military Misconduct and Homelessness Among US Veterans Separated From Active Duty, 2001-2012. JAMA. 2015 Aug 25; 314(8):832-4. [view]
13. Gilmore AK, Brignone E, Painter JM, Lehavot K, Fargo J, Suo Y, Simpson T, Carter ME, Blais RK, Gundlapalli AV. Military Sexual Trauma and Co-occurring Posttraumatic Stress Disorder, Depressive Disorders, and Substance Use Disorders among Returning Afghanistan and Iraq Veterans. Women's health issues : official publication of the Jacobs Institute of Women's Health. 2016 Sep 1; 26(5):546-54. [view]
14. Gundlapalli AV, Beekmann SE, Graham DR, Polgreen PM, Infectious Diseases Society of America's Emerging Infections Network. Perspectives and concerns regarding antimicrobial agent shortages among infectious disease specialists. Diagnostic Microbiology and Infectious Disease. 2013 Mar 1; 75(3):256-9. [view]
15. Divita G, Workman TE, Carter ME, Redd A, Samore MH, Gundlapalli AV. PlateRunner: A Search Engine to Identify EMR Boilerplates. Studies in health technology and informatics. 2016 Jan 1; 226:33-6. [view]
16. Toth DJ, Gundlapalli AV, Schell WA, Bulmahn K, Walton TE, Woods CW, Coghill C, Gallegos F, Samore MH, Adler FR. Quantitative models of the dose-response and time course of inhalational anthrax in humans. PLoS pathogens. 2013 Aug 1; 9(8):e1003555. [view]
17. Divita G, Shen S, Carter ME, Redd A, Forbush T, Palmer M, Samore MH, Gundlapalli AV. Recognizing Questions and Answers in EMR Templates Using Natural Language Processing. Studies in health technology and informatics. 2014 Jan 1; 202:149-52. [view]
18. Tran LT, Divita G, Redd A, Carter ME, Samore M, Gundlapalli AV. Scaling Out and Evaluation of OBSecAn, an Automated Section Annotator for Semi-Structured Clinical Documents, on a Large VA Clinical Corpus. AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2015 Nov 5; 2015:1204-13. [view]
19. Gundlapalli AV, Divita G, Carter ME, Redd A, Samore MH, Gupta K, Trautner B. Taming Big Data: An Information Extraction Strategy for Large Clinical Text Corpora. Studies in health technology and informatics. 2015 Jan 1; 213:175-8. [view]
20. Divita G, Carter ME, Tran LT, Redd D, Zeng QT, Duvall S, Samore MH, Gundlapalli AV. v3NLP Framework: Tools to Build Applications for Extracting Concepts from Clinical Text. EGEMS (Washington, DC). 2016 Aug 11; 4(3):1228. [view]
21. Gundlapalli AV, Redd A, Carter M, Divita G, Shen S, Palmer M, Samore MH. Validating a strategy for psychosocial phenotyping using a large corpus of clinical text. Journal of the American Medical Informatics Association : JAMIA. 2013 Dec 1; 20(e2):e355-64. [view]

Conference Presentations

1. Gundlapalli AV, Samore MH, Palmer M, Tuteja AK, Carter M, Shen S, South B, Forbush T, Divita G. Annotation of Symptoms in VA Clinical Documents. Poster session presented at: Integrating Data for Analysis, Anonymization, and Sharing Annual Conference; 2012 Sep 29; La Jolla, California. [view]
2. Gundlapalli AV, Samore MH, Nelson R, DuVall SL, Shen S, South B, Palmer M. Classifying Features from VA Clinical Documents to Identify Homeless or At-Risk Veterans. Poster session presented at: VA HSR&D National Meeting; 2011 Feb 16; National Harbor, MD. [view]
3. Gundlapalli AV, Samore MH, Nelson R, DuVall SL, South B, Shen S, Palmer M. Identification of features for detection and prediction of homelessness from VA clinical documents. Poster session presented at: International Society for Disease Surveillance Annual Conference; 2010 Nov 30; Park City, UT. [view]
4. Hope CJ, Garvin JH, Gundlapalli AV. Incomplete and selective Documentation of delirium in the VA Electronic medical Record. Poster session presented at: American Medical Informatics Association Annual Symposium; 2011 Oct 23; Washington, DC. [view]
5. Samore MH, Nelson R. Screening for Homelessness in the Free Text of VA Clinical Documents using Natural Language Processing. Poster session presented at: VA HSR&D / QUERI National Meeting; 2012 Jul 16; National Harbor, MD. [view]
6. Palmer M, Shen S, Divita G, Samore MH, Gundlapalli AV. Using clinical mental models to guide annotation of medically unexplained symptoms and syndromes found in VA clinical documents. Poster session presented at: International Society for Disease Surveillance Annual Conference; 2010 Dec 1; Park City, UT. [view]
7. South B, Palmer M, Shen S, Divita G, DuVall SL, Samore MH, Gundlapalli AV. Using Clinician Mental Models to Guide Annotation of Medically Unexplained Symptoms and Syndromes found in VA Clinical Documents. Paper presented at: International Society for Disease Surveillance Annual Conference; 2011 Dec 7; Park City, UT. [view]

DRA: Mental, Cognitive and Behavioral Disorders, Health Systems
DRE: Research Infrastructure, Epidemiology, Diagnosis
Keywords: none
MeSH Terms: none