Early screening for gastrointestinal diseases is of vital importance for reducing mortality through introducing early intervention. In this paper, a biomimetic artificial whisker-based hardware system with artificial intelligence-enabled self-learning capability is proposed for endoluminal diagnosis. The proposed method provides an end-to-end screening strategy based on tactile information to extract the structural and textural details of the tissues in the lumen, enabling objective screening and reducing the inter-endoscopist variability. Benchmark performance analysis of the proposed was conducted to assess the electrical characteristics and core functions. To validate the feasibility of the proposed for endoluminal diagnosis, an ex-vivo study was conducted to detect some common tissue structures and our method shows promising results with the test accuracy up to 94.44% with 0.9167 kappa. This previously unexplored tactile-based method could potentially enhance or complement the current endoluminal diagnosis.