VisualDx in Otolaryngology: Assessing the Performance of a Differential Diagnosis Tool

Introduction: VisualDx is a clinical decision-making support tool that generates differential diagnoses by synthesizing clinical data entered by the user. While VisualDx is frequently used in fields such as dermatology, its utility in otolaryngology remains unexplored. In this study, we assessed the effectiveness of VisualDx in accurately diagnosing otolaryngological complaints.

Methods: Patients presenting with the most common chief complaints at the otolaryngology clinic of a tertiary academic center (oral lesion, otalgia, pharyngitis, dysphagia, rhinorrhea, nasal congestion, and tinnitus) were retrospectively identified, and a chart review was conducted. Demographic information, past medical history, social history, symptoms, and physical exam findings were entered into VisualDx. Differential diagnoses generated by VisualDx were compared to the attending physician’s diagnosis. A case was considered a “success” when VisualDx correctly identified the attending diagnosis among the top five differential diagnoses, with its rank (1-5) based on its position on the list. Descriptive statistics and logistic regression analyses were conducted.

Results: This study included 140 patients (54.3% female, median age = 54 years). VisualDx identified the true diagnosis for 80 of the 140 patients (57.1%). The success rate of VisualDx was highest for patients with a chief complaint of rhinorrhea (85%) and lowest for those with otalgia (35%). Among successful cases, VisualDx most consistently ranked the true diagnosis as the top result for patients with tinnitus (90.9%). VisualDx was significantly more likely to correctly diagnose rhinorrhea compared to otalgia (OR: 10.99, 95% CI: 2.21–55.56, p=0.003) and oral lesions (OR: 6.25, 95% CI: 1.28–30.30, p=0.024).

Conclusions: VisualDx can provide clinical diagnostic support in otolaryngology, though its accuracy varies by presenting chief complaint. VisualDx may have utility in education and training by helping learners develop a framework for diagnostic reasoning and enhance pattern recognition for both common and uncommon presentations of clinical syndromes.