The use of natural language processing (NLP) could be a useful predictor of the openness of communication about genetic cancer risk in patients from families affected by hereditary breast and ovarian cancer (HBOC) syndrome, according to study findings published in the journal JMIR Formative Research.

The researchers explained that “information exchange about genetic cancer risk may be easier with some family members or may present a particularly difficult moment with others.”

NLP is an analytic technique for evaluating human language to extract information, to perform sentiment and semantic analysis, and for other purposes related to text analysis. Sentiment analysis is based on assessing a person’s opinions, emotions, and attitudes regarding a certain factor, such as an experience. NLP has been utilized in health care in evaluations of patient-reported outcomes, but prior to this study, the researchers indicated that the use of NLP and sentiment analysis had not been applied to communication within families regarding a genetic cancer risk.

Continue Reading

The researchers conducted this study in the context of a larger study involving the Swiss CASCADE cohort. Recruited patients had pathogenic variants related to HBOC or Lynch syndrome; 44 patients were evaluated from Switzerland, and 9 were based in South Korea.

Patients were evaluated by NLP analysis based on narrative data they provided about family communication of test results. Patients also completed surveys in which they provided responses to questions about demographic and clinical characteristics and various aspects of family communication involving genetic cancer risk. These surveys used Likert-type scales to indicate level of agreement with statements.

Based on analyses of narrative data and survey data, the researchers explored factors that may be linked to a higher level of “openness of communication” regarding genetic test results and cancer risk with both family and healthcare providers.

The researchers identified several factors that appeared to be associated with greater openness of communication. These included having a higher overall net sentiment score obtained in analysis of the patient narrative, as well as having higher fear, being single, having a nonacademic education, and having higher informational support at the family level. Patients on average reportedly showed an overall trend toward openness in communication, and the words “cancer,” “family,” and “time” were the 3 most common terms extracted from patient narratives.

The researchers considered the NLP approach used in this study to reliably predict openness of communication regarding genetic cancer risk in this population. This was based on an accuracy level of 0.72 obtained in validation of the model, with an area-under-the-curve score of 0.69, a specificity score of 0.62, and a sensitivity score of 0.83.

“Overall, this experimental analysis provides evidence that our approach is promising and can be further used in the field of technology-mediated communication and precision public health,” the researchers concluded in their report.


Baroutsou V, Gonzalez Pena RC, Schweighoffer R, et al; CASCADE Consortium. Predicting openness of communication in families with hereditary breast and ovarian cancer syndrome: natural language processing analysis. JMIR Form Res. 2023;7:e38399. doi:10.2196/38399