NUMERO: #20260422

Título: DEVELOPMENT AND FIELD EVALUATION OF A MOBILE-BASED AI APPLICATION (SMITA) FOR ORAL DISEASE SCREENING

Nome do Apresentador: LAVANYA CHANDRA

Categoria do Trabalho: Pôster - Investigação Original (IO)

Área Temática: Odontologia hospitalar

Resumo: Objective :
Integrating Artificial Intelligence into oral healthcare offers immense potential for early disease detection, particularly in resource limited settings. This study details the development and field testing of the Simple Mobile Interface Technology Application (SMITA) for scalable screening
Study Design :
Comprehensive workflow was built, involving large-scale intraoral image acquisition via mobile devices, followed by expert annotation of lesion features and clinical metadata. This curated dataset trained an AI model integrated into the SMITA Android application. Field evaluation involved 60 participants split between community and dental settings. Trained community health workers and dentists captured images following a standardized protocol to ensure uniformity.
Results :
Significant performance variations based on context was noted. In community settings, model showed high specificity (85.06%) and accuracy (83.75%) but moderate sensitivity (42.86%). Conversely, institutional dataset yielded perfect sensitivity (100%) but lower specificity (59.49%) and precision (3.03%), indicating high false-positive rate. These discrepancies suggest operator-dependent and environmental factors influence AI performance despite standardized protocols.
Conclusion :
SMITA proves promising, scalable screening tool across diverse environments. However, the performance variability highlights necessity for context-aware refinement and training on heterogeneous datasets to improve robustness and clinical utility in real-world translational applications.

Autor 1:  LAVANYA CHANDRA

E-mail 1:  [email protected]

Autor 2:   MADAN KUMAR

E-mail 2:  [email protected]

Autor 3 :  RANGANATHAN KANNAN

E-mail 3:  [email protected]