AI System for Tongue Analysis Based on TCM

Challenge: The client needed to automate traditional Chinese medicine (TCM) tongue analysis, which was manual, limited in scalability, and prone to variability due to unstructured data, small datasets, and subjective expert interpretations. Solution: A multi-stage AI pipeline was developed, integrating NLP for data structuring, a CNN-based architecture for symptom detection and syndrome classification, and transfer learning, deployed on AWS with human-in-the-loop validation. Impact: The system achieved 80-90% accuracy in predicting TCM syndromes, significantly reducing analysis time, increasing patient throughput, and establishing a scalable, stable foundation for future automation.