A Community-Based Usability Study of an AI-Enabled Oral Cancer Screening App Operated by Village Health Volunteers:...

Area of Science:

• Public Health
• Artificial Intelligence
• Mobile Health

Background:

• Oral cancer poses a significant public health challenge, particularly in low- and middle-income countries with limited access to specialist care.
• Mobile health (mHealth) technologies, enhanced by artificial intelligence (AI), offer a scalable solution for extending oral cancer screening services to underserved populations.
• Village health volunteers (VHVs) in Thailand play a crucial role in delivering preventive services and connecting rural communities with healthcare specialists.

Purpose of the Study:

• To describe the technical development of RiskOCA, a smartphone-based AI-assisted platform for oral cancer risk assessment.
• To evaluate the usability of the RiskOCA platform when implemented by VHVs in a rural Thai setting.

Main Methods:

• RiskOCA was built with a 3-tier architecture: a patient interface for risk profiling and imaging, an embedded deep learning engine (DeepLab v3+ with ResNet-50) for lesion analysis, and a specialist portal for expert review.
• The AI model was trained on 2226 annotated intraoral images and validated.
• Field testing involved screening 1242 adults (≥40 years) in rural Thailand, with usability assessed via a 25-item questionnaire completed by 250 VHVs.

Main Results:

• The AI model achieved a mean classification accuracy of 93.22% across three diagnostic categories.
• Usability evaluation revealed high user satisfaction, with an overall mean score of 4.17 out of 5.
• VHVs reported high satisfaction, particularly with the app's contribution to older adult surveillance.

Conclusions:

• RiskOCA exhibits robust technical performance and high user acceptance, confirming its feasibility for community-based oral cancer screening.
• The integration of AI-assisted triage and expert review holds potential to decrease diagnostic delays and broaden screening coverage.
• This platform serves as a scalable model for oral cancer prevention in resource-limited settings.