Evaluating the Performance of Mobile Machine-Learning Platforms for Syphilis Symptom Screening

Key Points

• The aim is to assess the effectiveness of machine-learning models in classifying syphilis cases using clinical images and meta-data.
• Evaluation of three machine-learning models
• Classifying 39 cases of primary and secondary syphilis
• Analysis based on associated meta-data and clinical images
• All three models classified 33 images correctly
• Achieved an overall percent agreement of 84.6%
• 95% Confidence Interval ranged from 69.5% to 94.1%