JMIR Publications

ABSTRACT

Background:

Culturally and linguistically diverse populations are often underrepresented in population-based research and surveillance efforts. Such underrepresentation may lead to biased study results, limiting their generalizability in practice. Oftentimes, these populations are considered to be so-called hard-to-reach groups in public health outreach and information campaigns. As a result, these groups are challenged by two effects: the medical and health knowledge is less tailored to their needs and, at the same time, it is less accessible for them. In many cases, the first barrier is language, that is, individuals receiving public health material or study invitations struggle to understand information presented in languages in which they are not proficient. Modern machine translation tools might offer a cost-effective solution to this problem.

Objective:

This scoping review aims to systematically investigate current use-cases of MT specific to the fields of PH and epidemiology, with a particular interest in its use for population-based recruitment methods.

Methods:

PubMed, PubMed Central, Scopus, ACM Digital Library, and IEEE Xplore were searched for articles reporting on the use of MT in PH and epidemiological research for this PRSIMA-ScR compliant scoping review (pre-registration: osf.io/avt73). Information on communication scenarios, study designs and principle findings for each article were mapped according to a settings approach, ‘WHO Monitoring & Evaluation Maturity’ and the 'Service Readiness Level' frameworks, respectively.

Results:

Of 3,896 articles identified, 46 studies were included in this review with the earliest one dating from 2009. Most studies discussed the application of MT to existing PH materials, limited to one-way communication between PH officials and addressed audiences. No specific article investigated the use of MT for recruiting linguistically diverse participants to population-based studies. Regarding study designs, nearly three quarters (n=34) provided technical assessments of MT from one language (mainly English) to a few others (e.g., Spanish, Chinese or French). Only a few (n=12) explored the end-user attitudes (mainly of PH employees), while none deepened on the legal or ethical implications of its use. Experiments primarily involved PH experts with language proficiencies. Overall, the majority of summarizing results (n=38/70 statements) presented mixed and inconclusive views on the technological readiness of MT for PH information.

Conclusions:

Utilizing MT in epidemiology and PH can enhance outreach to linguistically diverse populations. Translation quality of current commercial MT solutions (e.g., Google Translate, DeepL) is sufficient if post-editing is a mandatory step in the translation workflow. Post-editing of legally and/or ethically sensitive material requires staff with adequate content knowledge in addition to sufficient language skills. Unsupervised MT is generally not recommended. Research on whether machine-translated texts are received differently by addressees is lacking, as well as research on MT in communication scenarios that warrant a response from the addressees.