JMIR Publications

ABSTRACT

Background:

Health researchers are increasingly using Natural Language Processing (NLP) to study a variety of mental health conditions using both social media and Electronic Health Records (EHRs). There is currently no published synthesis which relates specifically to the use of NLP methods on bipolar disorder, and this scoping review was conducted to synthesize the valuable insights that have been presented in the literature.

Objective:

This scoping review explored how NLP methods have been used in research to better understand bipolar disorder, and identify opportunities for further use of these methods.

Methods:

This scoping review was conducted with reference to the framework proposed by Arksey and O'Malley, Levac et al. and Daudt et al., and also informed by the guidance provided in the JBI manual for evidence synthesis in scoping reviews. A systematic, computerized search of index and free-text terms related to bipolar disorder and NLP was conducted using five databases and one anthology: MEDLINE, PsycINFO, Academic Search Ultimate, Scopus, Web of Science Core Collection, and the ACL Anthology.

Results:

Of the 507 studies identified, 35 met inclusion criteria. A narrative synthesis was used to describe the data, and the studies were grouped into 4 objectives: (1) Prediction and classification (n=25), (2) Characterizing the language of bipolar disorder (n=13), (3) Using EHRs to measure health outcomes (n=3), and (4) Using EHRs for phenotyping (n=2) . Examples of the applications of NLP methods included predicting a diagnosis, predicting emotional state after online interaction, characterizing the linguistic features of people living with bipolar disorder, and performing phenotyping of bipolar disorder within health records. Ethical considerations were reported for 60% (21/35) of the studies.

Conclusions:

There are increasing opportunities for interaction between the clinical and natural language processing communities, and existing research conducted both on social media and EHR data demonstrates how the analysis of language can be used to assist and improve in the provision of care for people living with bipolar disorder. Under researched areas that could benefit from the use of NLP methods include research into risk-taking in bipolar disorder, understanding which online interactive services would be of greatest benefit to support those living with bipolar disorder, building a wider picture of social and occupational functioning in people living with bipolar disorder, and learning more about the representation of gender in bipolar disorder populations online. Future research which implements NLP methods to study bipolar disorder should be governed by ethical principles (which is not apparent for some of the papers included within this review) and any decisions regarding the collection and sharing of datasets should ultimately be made on a case by case basis considering the risk to the data subjects and being able to ensure their privacy.