Bao Do is an expert in radiology informatics, computer vision, and quantitative musculoskeletal imaging. He has developed and validated deep-learning models for diagnostic interpretation, hardware recognition, and automated reporting across orthopedic and radiographic domains. His recent studies demonstrated high-performance CNNs for detecting perilunate and lunate dislocations on wrist radiographs (AUC = 0.986) ?Pridgen et al., Plast Reconstr Surg 2023; 10.1097/PRS.0000000000010928? and improving clinician accuracy through machine-learning-assisted diagnosis in a multicenter reader study ?Luan et al., Hand (N Y)2025; 10.1177/15589447241308603?. He co-developed AI systems for automated classification of hip hardware achieving radiologist-level accuracy (AUC = 0.99) ?Ma et al., J Imaging Informat Med 2024; 10.1007/s10278-024-01263-y?, scoliosis curvature measurement from 2,150 spine radiographs ?Ha et al., J Digit Imaging 2022; 10.1007/s10278-022-00595-x?, and fully automated leg-length analysis and reporting ?Larson et al., J Digit Imaging2022; 10.1007/s10278-022-00671-2?. Earlier work included Bayesian models for bone tumor diagnosis ?Do et al., J Digit Imaging 2017; 30:709-13?, semantic content-based image retrieval using relevance feedback ?Banerjee et al., J Biomed Inform 2018; 84:123-35?, and NLP-based uncertainty detection in radiology reports ?Callen et al., J Digit Imaging 2020; 33:1209-19?, demonstrating a career-long commitment to explainable, data-driven imaging analytics.

Interests: Musculoskeletal imaging AI, AI for workflow optimization, human-AI interaction in radiology, scalable education

www.stanford.edu/~baodo