Print, check, repeat: digital quality by design for 3D-printed medicines using OpenAI models

Semi-solid extrusion three-dimensional printing (SSE-3DP) technology has emerged as a versatile manufacturing approach across a wide array of industries. Despite the rapid progress of additive manufacturing technologies and their increasing translation beyond laboratory environments, the absence of widely standardized quality control (QC) strategies remains a challenge to broader and routine implementation. Post-processing inspection remains challenging due to variability in material, geometry, and optical properties. This study aimed to establish a systematic and scalable framework for QC of SSE-printed patches by integrating colorimetric analysis with Artificial Intelligence (AI)-assisted machine vision. Three classes of hydrogel inks (starch-based, pectin-based, and gelatin-based) were printed as model patches on platforms of different colors. Perceptual color contrast between printed patches and printing platforms was quantified on-print using CIEDE2000 color difference metric and validated through image-based analysis. OpenAI tools (ChatGPT-5 (Plus) and Perplexity Pro) were then applied to support color prediction and selection, image segmentation, and geometric comparison between printed structures and their corresponding Computer-Aided Design (CAD) models using Intersection over Union (IoU) metrics. The results demonstrated that optimized color combinations significantly improved edge detection, segmentation accuracy, and defect identification, enabling reliable differentiation between well-printed and defective patches. Furthermore, the AI-assisted workflow successfully quantified infill accuracy and structural deviations without specialized machine vision hardware. Overall, this work develops a perception-driven, AI-powered QC strategy that improves reproducibility, enables automated inspection, and offers a practical path toward standardized quality assurance in SSE-3DP.