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

Area of Science:

• Medical Informatics
• Artificial Intelligence in Medicine
• Dermatology

Background:

• Syphilis diagnosis relies on clinical presentation and laboratory tests.
• Early detection of primary and secondary syphilis is crucial for effective treatment and public health.
• Machine learning offers potential for analyzing complex medical data, including clinical images.

Purpose of the Study:

• To evaluate the efficacy of machine-learning models in classifying syphilis cases.
• To assess the performance of AI in differentiating between primary and secondary syphilis using clinical data and images.
• To explore the potential of machine learning for patient-driven symptom screening.

Main Methods:

• Three distinct machine-learning models were developed and trained.
• Models utilized associated metadata and clinical images from 39 confirmed syphilis cases.
• Model performance was evaluated based on classification accuracy and agreement rates.

Main Results:

• All three machine-learning models achieved high classification accuracy.
• The models correctly classified 33 out of 39 syphilis cases.
• An overall percent agreement of 84.6% (95% CI 69.5-94.1%) was observed across the models.

Conclusions:

• Machine-learning models demonstrate significant potential for accurate syphilis classification.
• AI-powered tools can effectively analyze clinical images and metadata for disease identification.
• These models may facilitate patient-driven symptom screening, potentially improving early syphilis detection.