Modeling workflow to design machine translation applications for public health practice

Area of Science:

• Public Health Communication
• Health Informatics
• Machine Translation

Background:

• Limited English proficiency populations face barriers accessing health information.
• Translating health promotion materials is time-consuming and costly.
• Existing translation workflows in public health need optimization.

Purpose of the Study:

• To understand information workflow processes for translating health materials for limited English proficiency individuals.
• To inform the design of context-driven machine translation (MT) tools for public health (PH).

Main Methods:

• Applied a cognitive work analysis framework in two Washington State health departments.
• Conducted interviews and task analysis with public health professionals.
• Validated findings and modeled translation workflows to identify system requirements.

Main Results:

• Detailed description of the translation work in public health.
• Developed an information workflow diagram for health material translation.
• Identified themes and design implications for integrating MT in PH, including a prototype tool.

Conclusions:

• Understanding work context and information workflow is essential for designing effective systems.
• MT requires human review for quality assurance and practical adoption in public health.
• Design considerations for MT in PH include cost-effectiveness and quality concerns.