Application of ChatGPT as a support tool in the diagnosis and management of acute bacterial tonsillitis

Artificial intelligence (AI) is transforming medicine through techniques like machine learning and deep learning. AI aids diagnosis, enhances patient care, and streamlines healthcare systems. Despite potential benefits, challenges of bias and trust must be managed. Tonsillitis is a common otolaryngological condition with economic implications. The study assesses ChatGPT’s utility in the diagnosis management of bacterial tonsillitis, highlighting its potential for patient-professional interaction.

2 methods evaluated ChatGPT 3.5 diagnostic ability for tonsillitis: patient-written cases and specialist-created cases. The scenarios involved real patients and fictional cases, assessed by 15 otolaryngologists and 5 pediatricians. Variables included diagnosis accuracy, recommendations quality, and message count.

A total of 35 conversations were conducted. ChatGPT achieved accurate diagnoses in 100% of cases, with an average of 3.7 ± 1.1 chat entries for diagnosis. No significant difference existed between professional and patient scenarios (p = 0.977). Recommendations were categorized: appropriate (48.57%), incomplete (45.71%), inappropriate (5.71%), with no significant intergroup difference (p = 0.196). ChatGPT consistently advised consulting a doctor and exhibited expertise in guiding medical consultations.

ChatGPT demonstrates promise in providing medical insights and general advice. Its diagnostic accuracy for tonsillitis is notable, but it relies on static data and lacks individual history assessment. ChatGPT shows potential for diagnostics in simpler cases like tonsillitis, but accuracy for complex conditions needs refinement. Further research is needed for validation and broader application.