LingLan improves Traditional Chinese Medicine diagnosis using multimodal AI
LingLan: An Advancing Traditional Chinese Medicine Diagnosis LLM with Multimodal Data
Artificial Intelligence
Summary
Traditional Chinese Medicine (TCM) uses broad and subjective methods for diagnosis, making it hard to use modern AI techniques. The authors developed LingLan-14B, an AI model that learns TCM diagnostic steps by turning tongue and pulse images into structured digital records. Their system significantly improves diagnostic accuracy by combining information from multiple TCM methods. This advancement helps make TCM diagnosis more precise and consistent using AI.
What this means in practice
- •For tcm clinic teams: Automate and enhance TCM diagnostic processes by converting images and patient inquiries into standardized digital records for more accurate diagnoses.
- •For health ai developers: Incorporate structured multimodal TCM diagnostic data into large language models to improve AI’s understanding of holistic medical approaches.
Authors
Zheng Chen, Zhicheng Du, Haoxuan Li, Yingshan Liang, Peiwu Qin
Abstract
Though artificial intelligence (AI) increasingly transforms modern medicine, its integration into Traditional Chinese Medicine (TCM) has been relatively slow, primarily due to TCM's reliance on holistic, subjective diagnostic methods---namely Inspection, Auscultation and Olfaction, Inquiry, and Palpation(I-AOI-P)---which are difficult to align with quantitative, standardized medical systems. In this work, we introduce a Unification Framework for Multimodal Data (UFMD), which automatically processes tongue and pulse images into structured, clinically standard descriptions, integrating multi-source diagnostic information into a unified digital record of I-AOI-P process. Building on this structured data, we create LingLan-14B, a TCM-specific large language model fine-tuned via supervised learning to emulate the diagnostic logic and workflow of I-AOI-P process. Experimental results show that our method significantly enhances diagnostic accuracy, achieving a relative improvement of 103.5% over the baseline (62.72% vs. 30.82%) and reaching an F1-score of up to 82%.