Comparison of artificial intelligence generated visuals with visual analog scale for pain assessment

Traditional pain assessment tools such as the Visual Analog Scale (VAS) rely heavily on patients' cognitive ability to quantify pain, which may not effectively capture the complexity of the pain experience. This study investigates the use of artificial intelligence (AI)-generated visuals as an alternative method for postoperative pain assessment.

This prospective, single-center study enrolled 400 postoperative patients aged 18 years and older. Patients first evaluated their pain using VAS and then selected from five AI-generated images depicting various pain intensities. After both assessments, participants completed a survey comparing the two methods in terms of clarity, ease of use, and perceived usefulness.

Image-based assessment was preferred by 73.5% of participants, while 25.5% favored VAS (P=0.001). Paired t-tests showed that image-based assessment scored significantly higher for ease of interpretation (66.49±25.17 vs. 36.23±28.21), clarity (67.59±25.36 vs. 38.10±29.40), and usefulness (67.59±25.62 vs. 37.79±29.38), all with P<0.001. High image selection accuracy was observed for mild (VAS 1-2, 93.3%) and severe (VAS 9-10, 95.1%) pain levels, with moderate accuracy in mid-range scores.

AI-generated visuals offer a promising, patient-friendly alternative to traditional numeric pain scales. This novel approach demonstrated higher user satisfaction and more intuitive pain communication, particularly at pain extremes. Further refinement and validation are needed to optimize mid-range pain visuals and explore broader clinical applicability.