Traditional medicine AI translation engine oriented to education and scientific research

Abstract

The invention provides a traditional medicine AI translation engine, which converts literatures such as traditional Chinese medicine and Ayurbark from a concept space to a modern medicine semantic space. A neural machine translation and semantic mapping combined double-stage method is adopted, ancient Chinese, Sanskrit and the like are translated firstly, then a traditional concept is automatically annotated as a modern term (such as liver depression and qi stagnation)-liver dysfunction caused by emotion depression), and a knowledge graph associated entity is constructed. And multi-language and multi-format output is supported, and semantic annotation teaching materials or computable knowledge data are generated, so that non-traditional Chinese medicine background researchers can understand medical cases and take the medical cases into evidence-based research. International propagation and academic mining of traditional medicine are promoted, traditional Chinese medicine digitization is assisted, and great strategic significance is achieved.

Description

Technical Field

[0001] This invention relates to the fields of medical knowledge management and machine translation technology, specifically to an AI translation engine for traditional medicine aimed at education and research. This engine automatically converts classic texts of Traditional Chinese Medicine and Ayurvedic case studies into modern semantically structured versions, thereby promoting the understanding, teaching, and reuse of traditional medical knowledge within the global medical community. Background Technology

[0002] Traditional Chinese medicine classics (such as the *Huangdi Neijing* and *Shanghan Lun*) and other traditional medical texts contain rich experience and theories, but their obscure language and unique conceptual systems make them difficult for many modern researchers and students to accurately understand. A significant portion of this knowledge cannot be directly incorporated into the framework of modern medical education and research. For example, the phrase "the kidney governs water" in a classic TCM text encompasses the complex concepts of the kidneys and fluid metabolism, which is difficult for those unfamiliar with TCM theory to fully comprehend. Ayurvedic case records also frequently use Sanskrit terminology and traditional theories, differing greatly from the format of modern medical literature. In recent years, AI translation and knowledge graph technologies have developed rapidly, but general machine translation systems cannot handle these highly specialized texts that cross semantic systems. Some attempts exist, such as manually compiling TCM terminology collections and bilingual annotated classics, but these are limited in scale and time-consuming. How to leverage AI to digitize and structure traditional medical knowledge and express it in modern scientific language has become a crucial step for traditional medicine to reach a global audience. This will not only help researchers from different countries compare the effectiveness of traditional therapies but also contribute to cultivating TCM professionals with a global perspective. However, there is currently a lack of a dedicated intelligent translation engine that can handle both language translation and semantic conversion, translating ancient medical texts into modern, readable, and computable forms. Summary of the Invention

[0003] To address this, this invention proposes a traditional medicine AI translation engine, designed for educational and research needs, to achieve automatic semantic parsing and modern expression of traditional medical literature. The engine adheres to the principle of semantic sovereignty, ensuring that traditional medical concepts are not distorted or lost during translation, while simultaneously using the DIKWP model to map the textual meaning to the knowledge system of modern medicine. Specific functions include: translating passages from classic medical texts written in ancient Chinese, Sanskrit, etc., into modern language; annotating the medical concepts involved with their corresponding standard modern medical terms; and further outputting structured knowledge representations (such as knowledge graph triples, causal relationship graphs, etc.). For example, for the case "A patient suffers from liver qi stagnation, treated with methods to soothe the liver and relieve stagnation" from a traditional Chinese medicine text, the engine outputs the corresponding modern description: "The patient exhibits depression and liver-related symptoms, and the treatment involves antidepressants and methods to improve liver function," and represents this in structured data as "Depressive symptoms <- association -> abnormal liver function." For example, the Ayurvedic theory of "disorders of the three Doshas leading to disease" is translated and explained by the engine as: "Disorders of the three humors lead to disease, similar to endocrine and metabolic imbalances in Western medicine," and a corresponding relational diagram is provided. Through such translation and annotation, this engine allows researchers worldwide to understand its core content without needing to be proficient in the original classical language and theories, enabling them to reuse traditional knowledge in evidence-based evaluations and cross-system comparative studies. The key intellectual property value of this invention lies in constructing a unique mapping database and translation model from traditional medicine to modern semantics, which can serve as the foundation for digital educational content and research tools, possessing broad international application potential.

[0004] This invention includes the following technical solutions:

[0005] A multilingual classic corpus: This corpus collects and organizes classic texts from Traditional Chinese Medicine, Ayurveda, Islamic medicine, and other sources, along with their existing translations and annotations, to create a parallel corpus. For ancient Chinese medicine texts, the corpus pairs include the original text, modern Chinese translation, and English translation (if available); for Sanskrit and other texts, there are original-English translations. A terminology dictionary is also included, covering the corresponding explanations of key traditional medical terms and modern terminology.

[0006] Classical Language Processing Module: Develops word segmentation and syntactic analysis tools for Classical Chinese and Sanskrit. Utilizing pre-trained models (such as BERT for Classical Chinese and BPE for Sanskrit), it performs segmentation, part-of-speech tagging, and dependency analysis on classical language texts, enabling a syntactic understanding of subject-verb-object structures and modification relationships in Classical Chinese. This step lays the foundation for accurate translation.

[0007] Dual translation model: Employing a two-stage translation strategy.

[0008] Translation at the language level: Use neural machine translation (NMT) to translate traditional medical texts into modern languages (such as vernacular Chinese or English). The model is specifically constrained by a medical glossary to prioritize the selection of correct medical equivalents. For example, translate "spleen" as "spleen" instead of an incorrect general term.

[0009] Semantic layer conversion: Feed the preliminary translation result into the semantic parsing module. Combining the traditional-modern medical knowledge graph, map the concepts in the sentence to standard medical concepts. For example, recognize that "liver qi stagnation" corresponds to "depressive symptoms and liver function disorders", and add an explicit expression of this implicit information in the translated sentence. This stage ensures that the translation is not only a literal translation but also includes semantic correspondence through logical reasoning and rules.

[0010] Knowledge graph generation: Extract medical entities and relationships from the translated and parsed content to generate a structured knowledge graph or triples. The ontology basis adopted is a unified medical ontology similar to that of Invention 1 to connect with modern medical knowledge bases. For example, map "kidney" to the anatomical entity "kidney" and "water" to the physiological function "water metabolism" from the sentence "The kidney governs water". The relationship is "governs". Output triples: (kidney, governs, water metabolism).

[0011] Multi-mode output: The engine provides multiple output formats:

[0012] Annotated modern language text: Include explanations in the translation. For example, "The kidney governs water (The kidney dominates the body's water metabolism function)" for learners to read.

[0013] Concise dialogue version: Rewrite classical cases into modern case report forms, extracting elements such as background, symptoms, diagnosis, treatment, and results for medical researchers to quickly understand.

[0014] Structured data: Export knowledge data in JSON / XML or RDF formats for scientific research analysis or integration with other databases.

[0015] Interactive correction module: To continuously improve the translation quality, the system allows traditional Chinese medicine experts or language scholars to correct the translation results. The correction comparison will be fed back into the model training library to optimize the parameters of NMT and semantic mapping, making the engine translation gradually more refined and authoritative. Description of the drawings

[0016] Figure 1 It is a system flow block diagram of the traditional medical AI translation engine.

[0017] Figure 2 It is an example diagram of the engine translation output.

[0018] Figure 3 It is a schematic diagram of a fragment of the knowledge graph.

[0019] Figure 4 A schematic diagram of the user interface for correction. Detailed Implementation

[0020] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be emphasized that these embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the present invention.

[0021] Example 1: Translation and Annotation of a Classical Chinese Medicine Text. Taking the classic line from the *Huangdi Neijing Suwen* (The Yellow Emperor's Inner Classic - Basic Questions): "If one is harmed by wind in spring, diarrhea may occur in summer," as an example. The original sentence means that if one is harmed by wind in spring, diarrhea is likely to occur in summer. The engine first identifies the seasons "spring" and "summer" in classical Chinese, and the causes "wind" and symptoms "diarrhea." The NMT model translates this into modern English: "If one is harmed by wind in spring, diarrhea may occur in summer." Then, the semantic conversion module intervenes, recognizing that "wind" here is not ordinary wind but a concept of disease cause in traditional Chinese medicine, adding an annotation explaining it as "wind pathogen (a type of exogenous disease)." Simultaneously, "diarrhea" is labeled as "a type of diarrhea, possibly caused by disharmony of the spleen and stomach." The final annotated translation is: "If one is harmed by wind pathogen in spring, one may develop diarrhea (digestive upset) in summer."* A node is added to the knowledge graph: (Spring, Wind Pathogen, Summer Diarrhea). This result is provided to readers with a Western medical background, who can understand that the original text discusses the link between seasonal allergies / infections and summer diarrhea, which is similar to the pathogenic mechanism of climate factors in modern medicine.

[0022] Example 2: Translation of an Ayurvedic Case. A case recorded in Sanskrit describes a patient presenting with "indigestion and irritability due to excessive bile (Bile)." The engine translation renders the Sanskrit as "Excess bile leads to indigestion and irritability." The semantic module further interprets "bile" as an increase in Pitta in Ayurveda, associated with high gastric acid or bile reflux in modern medicine. Therefore, the annotation translates as: "Excess Pitta (bile, analogous to high gastric acid) leads to indigestion and irritability." The knowledge graph extracts: (excessive bile, leading to, indigestion) and associates it with the modern concept of "excessive gastric acid secretion." Researchers can see that ancient cases describing similar conditions of peptic ulcers or gastroesophageal reflux can serve as historical comparative studies of the evolution of these diseases.

[0023] Example 3: Multilingual Teaching Assistance. A TCM college's bilingual teaching required students to understand meridian theory. The teacher input the classic passage, "The collateral vessels of the Heart channel... its disease manifests as chest fullness in excess, and speechlessness in deficiency." The engine translated it into modern Chinese: "Diseases of the collateral vessels of the Heart channel, in excess, manifest as chest fullness; in deficiency, one cannot speak." Simultaneously, it translated it into English: "For the Heartchannel's collateral, in excess there is chest fullness, in deficiency one cannot speak." Semantic annotations added: "(Heart channel: anatomically corresponds to the cardiac neural network; 'speechlessness' may refer to aphasia due to stroke)." In this way, students quickly understood the possible modern correspondence of meridian pathogenesis (cardiac nerve disorders affecting the language center) by comparing the Chinese and English versions with annotations. This example demonstrates that engines can assist in the compilation and teaching of bilingual textbooks, accelerating the cultivation of internationally-minded talents who understand both traditional medical theory and modern medicine.

Claims

1. A traditional medical AI translation engine for education and research, comprising a corpus module, a classical language processing module, a dual translation model, a semantic conversion module, and a knowledge graph generation module, characterized in that: The corpus module stores the original texts of traditional medical literature such as TCM classics and Ayurvedic classics, along with corresponding modern language translation samples, forming parallel corpora and terminology dictionary resources. The classical language processing module uses a pre-trained language model to perform word segmentation and syntactic analysis on the original texts of traditional medicine, extracting key grammatical structures and medical terminology. The dual translation model includes sequentially connected language-level translation units and semantic-level conversion units. The language-level translation units translate the original text into modern expressions, while the semantic-level conversion units, combined with a pre-constructed traditional-modern medical concept mapping, replace or annotate the traditional medical concepts in the translation with corresponding modern medical terms or explanations. The knowledge graph generation module extracts medical entities and their relationships from the semantically converted text, maps them to a unified medical ontology, and forms a structured knowledge graph output.

2. The translation engine as described in claim 1, characterized in that: The language layer translation unit employs a neural machine translation model, incorporating medical terminology constraints and multilingual word alignment techniques during training to ensure that the translation faithfully reproduces the original meaning while using correct modern medical terminology. Preferably, important traditional medical terms are retained in the translation along with parenthetical explanations in the target language to balance preservation and understanding.

3. The translation engine as described in claim 1, characterized in that: The semantic layer conversion unit contains a hybrid engine based on rules and machine learning. For each medical concept in the translation result, it first looks up the corresponding modern concept in the terminology dictionary and determines the interpretation method. If there is a one-to-many possibility, it uses the context and knowledge graph relationship disambiguation algorithm to select the correct mapping. For implicit causal or mechanistic information, it provides supplementary explanations through pre-set classical knowledge rules, thereby achieving semantic equivalence conversion.

4. The translation engine as described in claim 1, characterized in that: It further includes a user correction module that allows experts with backgrounds in both traditional and modern medicine to review and edit the translation and semantic conversion results. The module compares the expert revisions with the original results and generates feedback signals to retrain the translation model or update the semantic conversion rules in order to continuously improve the accuracy of the engine.

5. The translation engine as described in claim 1, characterized in that: The engine supports multiple output formats, including annotated modern language translations, structured summaries of traditional cases, and visualizations based on the knowledge graph, enabling users to both read and understand the translated content and use the knowledge within for data analysis and scientific research comparisons.

Images