AI In Herbal Medicine

Artificial Intelligence (AI) is revolutionising Traditional Chinese Medicine (TCM) herbal practice by bringing precision, efficiency, and scientific rigour to its ancient foundations. Through machine learning, deep learning, and advanced data analytics, AI is enabling:

Herb Identification and Classification: AI recognises and classifies herbs based on chemical, botanical, and morphological features, improving quality control, standardisation, and authenticity verification.
Chemical Component Analysis: AI rapidly analyses complex herbal compounds, predicting pharmacological activity, toxicity, and interactions. It supports compound profiling, quality assessment, and ingredient discovery.
TCM Formula Discovery: AI models mine large TCM databases to identify novel herbal combinations, optimise classical formulas, and predict multi-target mechanisms using network pharmacology.
Personalised Prescriptions: AI analyses individual patient data—symptoms, constitution, lifestyle, and biometrics—to recommend tailored herbal prescriptions, enhancing efficacy and safety.
Drug Discovery and Development: AI accelerates the identification of bioactive compounds, predicts their targets, and models interactions. This supports high-throughput screening, and the discovery of new therapeutics derived from TCM.
Safety and Interactions: AI predicts herb-drug and herb-herb interactions (HDIs/DDIs), reducing the risk of adverse effects and improving integration with Western medicine.
Network Pharmacology: AI-powered network analysis reveals how herbs act across multiple biological targets and pathways, providing a systems-level view of their synergistic mechanisms and therapeutic potential.

These innovations are ushering in a new era of evidence-based, personalised, and globally relevant herbal medicine, aligning TCM with modern science while preserving its holistic principles. Through integrating AI and big data analytics, TCM is being modernised, enabling deeper insights into its foundational principles and bridging the gap between ancient wisdom and modern biomedical sciences. This fusion enhances the personalisation and precision of TCM-based therapies and sets the stage for a new era of holistic, data-driven healthcare.

Network Analysis

The application of Artificial Intelligence (AI) in network pharmacology significantly enhances our ability to decode the complex, multi-component, and multi-target nature of Traditional Chinese Medicine (TCM). One of the most transformative roles of AI lies in its ability to perform network analysis. This process maps and interprets the intricate interactions between herbal compounds, target genes, proteins, and metabolic pathways. AI leverages advanced algorithms and computational techniques such as:

Node Importance Evaluation: AI identifies key nodes—genes, proteins, or compounds—that play pivotal roles in disease modulation or treatment efficacy. These “hub nodes” are often critical targets for TCM compounds and may reveal the main therapeutic levers within a network.
Network Module Identification: AI isolates functional modules or sub-networks—clusters of tightly connected elements within the larger pharmacological network. These modules often represent biologically meaningful groupings such as pathways involved in inflammation, immune regulation, or metabolic balance, aligning closely with TCM treatment principles.
Synergistic Mechanism Discovery: By analysing the topological and functional relationships within the network, AI can uncover synergistic interactions between multiple herbs or compounds in a formula. This aligns with TCM's core concept of multi-component harmony, where the therapeutic effect arises from the balanced interplay of several ingredients, not a single active compound.

These AI-enhanced techniques support the construction of layered, dynamic pharmacological networks that map compound-target relationships and downstream effects on biological pathways and gene expression. By understanding how individual herbal components interact within the broader biological system, researchers can better predict therapeutic outcomes, minimise side effects, and refine TCM formulations for greater clinical efficacy.

In short, AI-driven network analysis transforms TCM from a traditionally empirical practice into a data-driven, systems-level science. It offers insights into the hidden architecture of healing mechanisms and bridges the gap between classical wisdom and modern biomedical research.

AI-Enhanced Network Pharmacology

Network pharmacology represents a paradigm shift in drug discovery and formulation development, especially within Traditional Chinese Medicine (TCM). It moves beyond the classical “single-drug, single-target” approach and embraces a “multi-component, multi-target” framework, which is inherently aligned with TCM's holistic nature. This systems-level perspective is particularly suited for treating complex, multifactorial diseases involving multiple physiological systems.

AI-enhanced network pharmacology brings powerful computational tools to this field, enabling the rapid analysis, prediction, and visualisation of compound-target-pathway-disease relationships. The Mechanism of AI-Enhanced Network Pharmacology in TCM includes:

Data Acquisition and Preprocessing: AI tools collect and clean high-volume data from TCM databases (e.g., TCMSP, TCMID, BATMAN-TCM), chemical structure databases, omics datasets (genomics, proteomics, metabolomics) and clinical records and biomedical literature. Natural language processing (NLP) extracts semantic and clinical meaning from unstructured texts, such as classical texts and case reports.

Compound Screening and Active Ingredient Identification: Machine learning algorithms (e.g., Random Forest, Support Vector Machines, Graph Neural Networks) predict and classify bioactive compounds based on ADME (absorption, distribution, metabolism, excretion) properties. Deep learning filters out ineffective compounds, identifying those with high drug-likeness and therapeutic relevance.

Target Prediction and Validation: AI models (e.g., DeepDTI, DeepConv-DTI) accurately predict compound-target interactions. AI–enhanced structure–activity relationship (SAR) models infer the biological relevance of molecular structures. Predictions are validated through molecular docking simulations and experimental data cross-referencing.

Construction of Compound-Target-Pathway-Disease Networks: Identified ingredients and targets are embedded into multi-layered pharmacological networks, where nodes represent compounds, genes/proteins, pathways, or diseases and edges represent known or predicted interactions. AI clustering and centrality analysis identify key hub targets, synergistic compound groups, and potential biomarkers.

Mechanism Elucidation and Therapeutic Optimisation: These networks are mined to reveal pathway enrichment (e.g., inflammation, apoptosis, immune regulation), mechanisms of synergy across herbs or ingredients and potential side effects or contraindication. AI evaluates combinations to recommend synergistic multi-herb formulations that optimise efficacy and safety.

Feedback Loop for Iterative Refinement: Clinical and experimental feedback is fed back into the AI model to refine predictions, improve robustness of the pharmacological network and personalise treatment recommendations based on patient-specific data.

A typical AI-enhanced network pharmacology workflow in TCM might involve selecting a classic TCM formula for rheumatoid arthritis, using AI to screen and predict the active ingredients in each herb, mapping each compound to immune-regulation pathways (e.g., IL-6, TNF-α), identifying shared targets between herbs that enhance anti-inflammatory effects and recommending formula modifications to reduce toxicity while preserving efficacy.

AI-enhanced network pharmacology is revolutionising TCM research by enabling the systematic elucidation of therapeutic mechanisms, accelerating multi-target drug discovery, and paving the way for precision herbal formulations. This integration makes TCM increasingly evidence-based, mechanistically understood, and globally scalable.

Hybrid Decision Support System

Martins et al. developed a novel hybrid decision support system (DSS) specifically designed to identify potential herb-drug interactions (HDIs) by integrating artificial intelligence (AI) technologies with pharmacological and botanical data. This system represents a significant step toward enhancing patient safety in integrative medicine, where Traditional Chinese Medicine (TCM) and Western pharmaceuticals are often used concurrently.

The hybrid DSS leverages machine learning algorithms and knowledge-based inference engines to predict new and previously unreported HDIs, cross-reference known pharmacokinetic and pharmacodynamic pathways and assess potential risk levels based on dosage, compound interaction profiles, and patient variables. By analysing vast datasets of clinical cases, pharmacological records, and known toxicological patterns, the system supports clinicians, pharmacists, and herbal medicine practitioners in making better-informed and safer treatment decisions. It not only reduces the risk of adverse drug events (ADEs) but also helps tailor therapy by accounting for individual patient profiles and known compatibilities.

This AI-enhanced approach brings greater precision, speed, and adaptability to a field traditionally reliant on empirical knowledge and manual cross-checking. As the integration of AI into TCM advances, systems like the one proposed by Martins et al. pave the way for a new era of intelligent, data-driven clinical decision support, ensuring the safe and effective co-use of herbs and conventional medicines.

Drug Discovery and Development in TCM

Traditional Chinese Medicine (TCM) represents a vast reservoir of natural compounds with therapeutic potential. From thousands of clinical uses, TCM includes many botanical, mineral, and animal-derived ingredients that inspire modern pharmacological research. Between 1981 and 2019, more than 60% of FDA-approved small-molecule drugs were either derived from or inspired by natural products, underscoring the relevance of TCM as a valuable source for drug discovery (Song, Chen, & Chen, 2024).

As the complexity of modern diseases increases, there is a growing demand for novel therapeutic agents with multitarget actions and lower side effect profiles—characteristics often found in traditional herbal formulations. Artificial intelligence (AI) is revolutionising this landscape by enhancing how bioactive compounds are identified, characterised, and translated into viable clinical treatments.

Accelerated Compound Screening and Identification: AI can process massive chemical datasets derived from TCM herbs, particularly through deep learning and machine learning algorithms. These models rapidly analyse chemical structures and identify novel active compounds, classify herbal constituents based on therapeutic potential, and reduce the time and cost of early-stage research by automating the screening process. Using high-throughput virtual screening (HTVS), AI can evaluate thousands of molecules against disease-specific targets in silico, prioritising those with the highest likelihood of efficacy for experimental validation.

Mechanism Elucidation and Target Prediction: AI models play a key role in predicting the molecular targets of TCM components. This involves mapping compound-target-pathway networks, simulating ligand–receptor interactions, and identifying multi-target synergies—a hallmark of TCM formulations. These capabilities help researchers uncover the pharmacodynamic mechanisms behind traditional prescriptions, offering a bridge between empirical knowledge and scientific validation.

Validation and Pharmacological Modelling: Once potential targets are identified, AI assists in bioinformatics analysis to confirm the biological relevance of targets, omics data integration (genomics, proteomics, metabolomics) to understand compound impact at a systems level and supporting experimental studies by narrowing focus to the most promising compound–target interactions. This data-driven prioritisation enhances efficiency and resource allocation in wet-lab experiments.

Predictive Toxicology and Safety Profiling: AI algorithms can also predict toxicity profiles of novel compounds, herb-drug interactions and potential adverse effects based on structure-activity relationships (SAR). This is especially important in TCM, where complex multi-herb formulas may pose risks if not accurately assessed. Early detection of potential safety concerns reduces the likelihood of clinical trial failures and improves patient safety.

Integration with TCM Databases and Knowledge Graphs: Advanced AI systems increasingly leverage knowledge graphs and semantic networks built from curated TCM databases. These platforms synthesise data from classical texts, pharmacopoeias, clinical case reports and experimental pharmacology. Integrating traditional wisdom with modern datasets allows AI to propose novel combinations or scientifically grounded reformulations consistent with TCM principles.

By combining ancient herbal wisdom with cutting-edge technology, AI is unlocking the full potential of TCM in drug discovery. It enables a systematic, data-driven, and personalised approach to identifying, evaluating, and validating herbal compounds for modern clinical use. This transformation not only accelerates the development of new, safe, and effective medicines but also reinforces the scientific credibility of TCM in global pharmaceutical research. As the field evolves, the integration of AI into TCM drug discovery promises to deliver innovative therapeutic solutions aligned with the holistic and preventative ethos of traditional Chinese medicine.

As artificial intelligence (AI) technology continues to evolve, its integration with herbal medicine marks the beginning of a transformative era that bridges the profound wisdom of Traditional Chinese Medicine (TCM) with the precision and capabilities of modern science. AI is emerging as a powerful catalyst in the research, development, and clinical application of herbal therapies, particularly in the following domains:

Chemical component analysis
Pharmacological mechanism exploration
New drug discovery
Quantitative and qualitative herb analysis
Herb classification and identification
Chemical profiling and standardisation

AI's unparalleled ability to process vast and complex datasets, uncover hidden patterns, and generate predictive models greatly surpasses the limitations of traditional research methods. This synergy between AI and herbal medicine unlocks deeper therapeutic insights and redefines how herbal knowledge can be applied in modern healthcare.

The complexity of herbal medicine—characterised by multifaceted interactions among numerous bioactive compounds—makes it a natural candidate for AI-enhanced analysis. Each herbal formula involves a dynamic interplay of ingredients that affect the body in intricate and often nonlinear ways. Understanding and optimising these interactions requires tools capable of high-level data integration and pattern recognition—capabilities at the core of AI systems.

AI enhances the accuracy, consistency, and efficacy of herbal diagnostics and treatments by adopting systematic, data-driven, and highly precise methodologies. More than just a technological advancement, this integration represents a paradigm shift: a movement toward intelligent herbal medicine that respects traditional foundations while embracing the innovations of the digital age.

AI In Herbal Medicine

The Identification, Classification and Recognition of Herbs

Artificial Intelligence (AI) is playing a transformative role in the identification and classification of herbal medicines by enabling the precise analysis of their chemical, physical, and pharmacological properties. Machine learning and deep learning technologies, in particular, are highly effective in analysing complex, high-dimensional data, making them ideal for interpreting the intricate chemical profiles of herbs.

These technologies facilitate the detailed characterisation of herbal components, helping researchers better understand their therapeutic effects, mechanisms of action, and potential interactions. This supports the discovery of novel applications for traditional herbs and enhances the standardisation, authentication, and quality control of herbal products—a crucial step toward ensuring consistency, safety, and regulatory compliance in clinical practice.

AI systems also address one of the greatest challenges in herbal medicine research: managing the vast and heterogeneous data generated from centuries of empirical knowledge, modern pharmacological studies, and chemical profiling. By applying data mining and network pharmacology approaches, AI helps reveal patterns and relationships within complex herbal formulations, offering insights into herb–herb interactions and the synergistic effects that underpin many traditional prescriptions.

Furthermore, AI excels in automated herb classification by analysing morphological features (e.g., leaf shape, colour, texture) and molecular fingerprints. Image recognition systems can identify herbal species with remarkable accuracy, while chemical analysis supported by AI can classify herbs based on active constituents, toxicity, and pharmacodynamic properties. Ultimately, AI is accelerating the scientific validation of herbal medicine and paving the way for a more standardised and mechanistic understanding of how traditional remedies work, making them more accessible, evidence-based, and globally integrated in modern healthcare systems.

Herbal Screening

AI plays a pivotal role in herbal screening by dramatically enhancing the speed, precision, and scope of identifying bioactive compounds in herbs and predicting their therapeutic effects. Here’s how AI works across the herbal screening process:

Data Collection and Integration: AI begins by processing large datasets that include traditional knowledge, pharmacological studies, chemical profiles, and genomic data of medicinal plants. Sources include classical TCM literature, pharmacopoeias, clinical trial data, high-throughput screening outputs and molecular databases (e.g., PubChem, TCMID, BATMAN-TCM). AI systems integrate these heterogeneous datasets to create a comprehensive knowledge base for herbal screening.

Feature Extraction and Compound Analysis: Machine learning (ML) algorithms are used to extract key features from herbal data including chemical structure (molecular fingerprints, SMILES notation), pharmacokinetics (ADME properties: absorption, distribution, metabolism, excretion), toxicity profiles and target prediction (which proteins or receptors a compound may bind to). Techniques like quantitative structure–activity relationship (QSAR) modelling and deep learning can predict biological activity based on chemical properties.

Bioactivity Prediction: AI models screen thousands of herbal compounds against known disease targets using supervised learning (e.g., SVM, random forests) to classify compounds as active/inactive, neural networks to find non-obvious patterns between structure and efficacy and docking simulations with AI-enhanced scoring systems to predict how strongly compounds bind to molecular targets. This reduces the time and cost traditionally associated with wet-lab bioassays.

Network Pharmacology and Systems Mapping: AI maps out the multi-target nature of herbal compounds by constructing herb–compound–target–disease (H-C-T-D) networks, pathway enrichment analyses showing how compounds influence biological systems and mechanism-of-action models, revealing how complex herbal formulations exert therapeutic effects. This aligns with the holistic philosophy of TCM, where herbs affect multiple pathways rather than a single target.

Compound Prioritisation: AI ranks compounds based on predicted efficacy, safety, drug-likeness and synergistic potential with other herbs. This helps researchers focus on the most promising bioactive candidates for further testing or drug development.

Continuous Learning: As new experimental data becomes available, AI models are retrained and refined, improving prediction accuracy over time. This iterative loop ensures the herbal screening process becomes more intelligent and effective with use.

AI revolutionises herbal screening by combining traditional knowledge with modern data science. This allows faster identification of promising herbal compounds, a deeper understanding of pharmacological effects, and safer and more effective integration of herbal treatments into contemporary medicine. This accelerates herbal drug discovery and modernises and validates centuries of TCM wisdom through scientifically rigorous methods.

TCM Formula Discovery

AI is transformative in discovering traditional Chinese medicine (TCM) formulas, helping modernise and accelerate what was once largely an experience-based and intuitive process. By analysing vast datasets, uncovering hidden patterns, and modelling complex herbal interactions, AI enables researchers and practitioners to identify, optimise, and design new herbal formulas with greater precision, safety, and scientific validity. Here’s how AI contributes to TCM formula discovery:

Mining Classical Texts and Clinical Records: AI leverages natural language processing (NLP) to extract structured knowledge from classical TCM literature, ancient prescriptions, and modern clinical case records. This includes identifying commonly used herbal combinations for specific patterns or conditions, extracting dosage patterns, contraindications, and treatment outcomes and structuring unstandardised textual information into searchable, analysable databases. By doing so, AI helps preserve and revitalise traditional knowledge, making it more accessible and actionable for modern research.

Herb Combination Prediction and Optimisation: AI models—especially association rule mining and clustering algorithms—analyse historical prescriptions to identify core formula structures (i.e., key herb combinations that frequently appear together), modular patterns based on syndrome types or organ systems and synergistic or antagonistic herb relationships, guiding safer and more effective combinations. This aids the refinement of existing formulas and the generation of new candidate formulas by recombining herbs with shared pharmacological or therapeutic profiles.

Network Pharmacology and Multi-Target Mapping: AI enables multidimensional mapping of herb–compound–target–disease relationships by integrating the chemical composition of herbs, biological targets (e.g., enzymes, receptors, genes), and disease-related pathways. Through network pharmacology, AI helps identify which combinations of herbs affect common molecular pathways, aligning with TCM’s holistic, multi-target therapeutic philosophy.

Personalised Formula Design: With access to individual patient data, such as tongue/pulse diagnosis, symptoms, lab results, and genetic markers, AI can tailor herbal prescriptions to the person’s unique constitution and condition. Algorithms evaluate herb suitability based on pharmacogenomics and allergies, historical outcomes for patients with similar profiles and recommended modifications to classical formulas to suit modern presentations. This enables the development of personalised herbal prescriptions aligned with TCM principles and precision medicine.

Virtual Screening and Formula Simulation: AI simulates how potential formulas behave in the body using in silico pharmacokinetic modelling (absorption, distribution, metabolism, excretion), toxicity screening, herb–drug interaction prediction, and target binding affinity simulation, particularly for chronic or systemic diseases. These simulations reduce trial-and-error in clinical practice and help prioritise formulas for experimental validation or clinical trials.

Formula Validation and Efficacy Prediction: AI tools like supervised machine learning models (e.g., random forests, support vector machines) analyse past clinical outcomes to predict the efficacy of candidate formulas, identify which patient populations are most likely to benefit and recommend dosage adjustments or co-therapies. This enhances evidence-based validation of new or modified TCM formulas.

AI supports TCM formula discovery by mining and structuring historical and clinical data, discovering new herb combinations through pattern recognition, mapping multi-target therapeutic pathways, designing personalised, effective prescriptions, simulating pharmacokinetics, and predicting efficacy. By combining classical wisdom with cutting-edge computational tools, AI paves the way for the next generation of TCM formulas that are scientifically grounded, patient-specific, and globally applicable.

Chemical Analysis

Artificial Intelligence (AI) is revolutionising chemical analysis within Traditional Chinese Medicine (TCM), driving advancements in drug discovery, ingredient profiling, and understanding pharmacological mechanisms. The inherent complexity of TCM formulas—often comprising multiple bioactive compounds with synergistic effects—necessitates sophisticated analytical tools. AI is uniquely equipped to address this challenge, especially through machine learning (ML) and deep learning (DL) algorithms.

Big Data Processing and Pattern Discovery: AI models process vast and diverse datasets from TCM chemical databases, enabling complex querying of herbal compounds, data visualisation of molecular interactions and interactive analysis platforms for researchers and clinicians. Natural Language Processing (NLP) is also employed to extract chemical and pharmacological information from classical texts and modern literature, turning unstructured data into structured, actionable knowledge.

Predicting Pharmacological Activity: Machine learning algorithms are trained to predict the biological activity of compounds based on chemical structure, functional groups, and molecular fingerprints. This helps discover new therapeutic agents, understand the mechanisms behind multi-component TCM formulas and identify active ingredients responsible for clinical efficacy. Such models refine TCM understanding from empirical observation to evidence-based pharmacology.

Mapping Compound-Biological Relationships: AI tools enable researchers to establish compound-target-pathway relationships by integrating chemical, genomic, and clinical data. Empowered by AI, network pharmacology models clarify how different compounds in herbal mixtures influence multiple biological targets, reflecting the holistic and multi-targeted nature of TCM.

Drug Interaction and Safety Prediction: AI predicts drug-drug interactions (DDIs), including interactions between TCM herbs and pharmaceuticals. AI can model drug-drug and herb-drug interactions, drug-food and drug-microbiome effects, and genetic factors affecting metabolism using large datasets and deep learning algorithms. This is particularly important in TCM, where complex polyherbal prescriptions can pose interaction risks. AI mitigates these risks by predicting incompatibilities, enhancing safety and treatment outcomes.

Enhanced Drug Discovery: AI accelerates the screening and development of novel compounds derived from TCM by identifying correlations between chemical structure and bioactivity. These models suggest promising lead compounds for further pharmacological testing, reducing the time and cost of traditional drug discovery.

Model Fusion for Deeper Insight: Combining AI methods, such as convolutional neural networks (CNNs), decision trees, and ensemble models, creates more robust, accurate frameworks for analysing chemical interactions in TCM. These hybrid approaches allow for better handling of noisy, imbalanced, or incomplete data, often found in herbal medicine research.

AI is reshaping TCM's chemical analysis landscape by making it more data-driven, predictive, and mechanistically informed. Its ability to model complex chemical interactions, predict therapeutic efficacy, and identify potential safety issues is essential in modernising herbal medicine. Ultimately, AI empowers TCM to move from traditional formulations to precision phytotherapy, where treatments are safer, more effective, and tailored to individual patient needs.

FordNet

FordNet is an advanced AI-based platform designed to enhance the accuracy and efficacy of Traditional Chinese Medicine (TCM) formulations. By integrating phenotypic (observable traits and treatment outcomes) and molecular (chemical and biological) information, FordNet provides intelligent, data-driven recommendations for herbal combinations. This significantly improves formulation hit rates, optimises compatibility, and increases the precision of therapeutic targeting.

At the core of FordNet is its ability to combine cutting-edge analytical techniques, such as liquid chromatography and mass spectrometry, with AI-powered data processing. This fusion enables the system to:

Characterise herbal compounds at a detailed molecular level.
Map bioactive compounds to specific phenotypic outcomes.
Predict synergistic effects between herbal components.
Refine dosage and compatibility for individualised prescriptions.

FordNet provides a comprehensive chemical profile of herbal formulations through this integrated approach, uncovering the pharmacological basis of TCM therapies. It improves quality control by identifying and quantifying active ingredients and ensuring that formulations are traditionally valid and scientifically substantiated. Additionally, FordNet supports formulation optimisation by suggesting modifications based on patient response patterns and real-time clinical feedback. This aligns with the broader trend of personalised medicine in TCM, where treatments are increasingly tailored to an individual’s constitution, symptoms, and underlying imbalances.

FordNet exemplifies how AI technologies can revolutionise TCM by transforming empirical knowledge into quantifiable, evidence-based practices. It bridges traditional wisdom with modern analytical science, enabling practitioners to deliver safer, more effective, highly personalised herbal treatments grounded in classical theory and molecular precision.

Data Fusion Techniques in TCM Research

Data fusion techniques are crucial in advancing Traditional Chinese Medicine (TCM) research. They integrate information from multiple analytical sources—such as spectroscopy, chromatography, mass spectrometry, and metabolomics—to construct a more holistic and accurate understanding of TCM formulations.

Rather than relying on a single dataset or method, data fusion combines complementary datasets at various levels:

Low-level fusion merges raw data from different instruments.
Mid-level fusion integrates features or patterns extracted from each dataset.
High-level fusion combines decision outputs or interpretation results to strengthen final predictions.

By synthesising diverse types of chemical, biological, and clinical data, this approach enables researchers to identify complex interactions between multiple herbal compounds, detect bioactive constituents with greater sensitivity and specificity, predict synergistic effects or potential contraindications and enhance the mechanistic understanding of traditional formulations.

In AI and machine learning, data fusion significantly improves model training and predictive accuracy, creating more robust algorithms for drug discovery, syndrome differentiation, and personalised treatment recommendations. Ultimately, data fusion promotes a systems-level approach to TCM—mirroring the holistic philosophy of Chinese medicine—while offering a rigorous scientific foundation for modern research, innovation, and global integration.

AlphaFold 3

AlphaFold3 is the latest advancement in protein structure prediction developed by DeepMind in collaboration with Isomorphic Labs. Building upon the revolutionary success of AlphaFold2, which accurately predicted 3D protein structures from amino acid sequences, AlphaFold3 extends these capabilities even further. It not only model’s proteins but also protein–nucleic acid, protein–ligand, and protein–ion interactions, offering a holistic view of complex biomolecular systems. This marks a major leap in the precision and scope of computational biology.

Integrating AlphaFold3 into TCM research is poised to accelerate the modernisation of herbal medicine and drug discovery by offering detailed insights into how natural compounds interact with biological targets at the molecular level.

Accelerating Drug Discovery from Herbal Compounds: TCM formulations often involve bioactive compounds whose mechanisms of action are not fully understood. AlphaFold3 enables researchers to predict the protein targets of herbal ingredients, model ligand–protein interactions between TCM compounds and human proteins, and identify multi-target binding profiles, aligning with the holistic, multi-pathway nature of TCM. This supports the rational development of TCM-based drugs with improved precision, efficacy, and safety.

Optimising Formula Design and Synergistic Combinations: Many TCM formulas involve multiple herbs working synergistically. AlphaFold3 allows for in silico simulation of compound-compound-protein networks, identification of synergistic effects based on shared or complementary protein targets and elimination of potential antagonistic interactions at the molecular level. This enhances formula refinement, leading to more effective and scientifically validated treatments.

Elucidating Mechanisms of Action: One of the challenges in integrating TCM with modern science has been the lack of mechanistic clarity. AlphaFold3 provides high-resolution predictions of how specific herbal compounds interact with disease-related proteins, which modulate metabolic, inflammatory, or immunological pathways and how personalised treatment might be adjusted based on an individual’s molecular profile. This bridges the gap between traditional empirical knowledge and modern biochemical understanding.

Supporting Safety and Toxicology Studies: Predicting off-target effects and potential toxicities is essential for clinical application. AlphaFold3 can identify unintended protein binding by TCM compounds, support AI-driven screening for toxicity, allergenicity, and cross-reactivity and enhance the safety profile of complex herbal formulas.

AlphaFold3 represents a groundbreaking tool in translational TCM research, offering molecular-level insights that align well with traditional herbal therapies' multi-component, multi-target nature. By integrating this technology into TCM formulas' research, validation, and optimisation, researchers can modernise practices without losing their holistic roots. Its application marks a new era where ancient wisdom meets molecular precision, allowing TCM to participate more fully in global biomedical innovation and evidence-based drug development.

Toxicity and Side Effects

Artificial Intelligence (AI) plays a pivotal role in enhancing the safety profile of Traditional Chinese Medicine (TCM) by predicting toxicity and adverse effects of herbal components. One of the major challenges in TCM drug development lies in the complex interactions among the multi-component formulas, where unintended side effects or toxic compounds may go undetected using traditional methods alone.

AI models—particularly machine learning (ML) and deep learning (DL) architectures—can process vast, multidimensional datasets from clinical records, pharmacological studies, molecular databases, and toxicological research. These models are trained to identify patterns and correlations between chemical structures and known toxic outcomes, allowing them to:

Predict potential toxicological risks before clinical trials.
Flag herb-drug interactions in polypharmacy settings.
Evaluate the dose-dependent safety profiles of individual herbs or formulas.
Recommend safer substitutes or adjustments in formulation.

By simulating thousands of interactions, AI helps practitioners and researchers avoid late-stage failures in drug development and improve clinical decision-making. These predictive insights are especially valuable for vulnerable populations such as the elderly, pregnant individuals, or those with chronic illnesses who may be more susceptible to adverse effects. Ultimately, AI-driven toxicity prediction advances the modernisation of TCM by ensuring that safety assessments are evidence-based, proactive, and scalable, aligning TCM practices with global pharmacovigilance and regulatory standards.

Drug-Drug Interactions

Drug-drug interactions (DDIs) are vital considerations in clinical practice, ensuring patient safety when multiple substances are used concurrently. In Traditional Chinese Medicine (TCM), herbal formulations often consist of complex combinations of various ingredients, and the risk of interactions, either with other herbs or with pharmaceutical drugs, is significantly heightened. The TCM principle of "incompatibility" (面相) has long acknowledged that certain herbal combinations may lead to adverse effects or diminished efficacy, necessitating a more precise and predictive approach to formulation.

As TCM becomes increasingly integrated with modern healthcare systems, understanding and predicting DDIs have become essential to mitigate potential risks. However, herbal medicine's holistic and synergistic nature presents challenges for conventional reductionist analysis. Modern technologies, particularly AI-enabled systems pharmacology, offer transformative solutions. These tools identify bioactive compounds within herbal formulations, predict molecular targets and pathways, clarify interaction mechanisms with pharmaceutical agents or other herbs and model complex networks of interactions using computational simulations. By integrating network pharmacology, machine learning, and large-scale molecular databases, AI systems can uncover known and previously hidden DDIs, offering a systematic, data-driven approach to TCM safety evaluation. These models simulate interactions in silico before clinical use, allowing for risk stratification of herbal combinations, optimised prescription recommendations and alerts for contraindications or potential toxicities.

This approach is especially valuable in personalised medicine, where patients may take herbal and Western pharmaceuticals. AI-enabled DDI prediction ensures greater clinical precision and fewer adverse reactions and contributes to TCM practices' global modernisation and regulatory integration.

Herbal-Drug Interactions (HDI)

As Traditional Chinese Medicine (TCM) becomes increasingly integrated with Western biomedical practices, the complexity of herbal-drug interactions (HDIs) emerges as a critical concern. TCM formulations typically comprise multiple herbs, each containing numerous active compounds that can interact unpredictably with synthetic pharmaceutical drugs. These interactions may enhance or inhibit drug efficacy or trigger harmful side effects. The intricacy and variability of such interactions have traditionally posed a significant barrier to safe and effective integrative healthcare.

Recent advancements in AI and the development of comprehensive HDI databases are helping to overcome these challenges. Machine learning algorithms—especially those trained on large datasets of clinical records, chemical structures, metabolic pathways, and pharmacokinetic profiles—can predict potential herb-drug interactions with increasing accuracy. These AI-powered models:

Analyse patterns in known HDI cases
Simulate chemical and biological interactions.
Forecast risks of toxicity, reduced efficacy, or synergistic effects.
Support clinical decision-making for safer prescription practices.

Compiling structured databases—integrating TCM pharmacopoeias, pharmaceutical drug libraries, chemical component libraries, and biomedical pathway maps—provides the foundational data infrastructure. These resources allow AI systems to cross-reference traditional herbal knowledge with modern pharmacological science, enhancing diagnostic precision and patient safety.

AI-based HDI models enable real-time interaction screening, customised alerts, and automated compatibility checks. They are essential tools for practitioners operating at the intersection of Chinese and Western medicine. They pave the way for a more predictive, personalised, and safe approach to integrative healthcare.

The PHYDGI Database

The PHYDGI (Phytochemical–Drug–Gene Interaction) Database is a specialised resource to address the increasing need for evidence-based herbal-drug interactions (HDIs) management. It integrates complex HDI data into a well-structured, scientifically validated format with citations to peer-reviewed studies and pharmacological literature. This database catalogues a wide array of herbal entries, each annotated with:

Interaction strength levels (e.g. mild, moderate, severe)
Mechanistic classifications (e.g. pharmacokinetic vs. pharmacodynamic)
CYP enzyme involvement, transporter systems, and receptor pathways
Associated clinical risks and documented outcomes.

By consolidating this information into an accessible platform, PHYDGI empowers stakeholders—including health practitioners, TCM clinicians, naturopaths, botanical medicine specialists, and food supplement manufacturers—to make more informed and safer prescribing decisions. Crucially, PHYDGI enhances safety by identifying and mitigating high-risk herb-drug combinations, supporting the development of safer botanical formulations, and promoting the rational, evidence-informed integration of TCM herbs and Western pharmaceuticals.

In the broader context of AI-assisted integrative medicine, PHYDGI serves as a core database that can be linked with machine learning models to enhance the prediction of adverse interactions and assist in real-time clinical decision-making. It represents a significant step toward standardising herbal-drug interaction data, advancing patient safety, and supporting the global modernisation of TCM and complementary therapies.

Drug Target's Networks

Drug-target networks are critical to understanding how TCM compounds interact with biological systems. These networks map the interactions between bioactive components in herbal formulations and their corresponding molecular targets, such as enzymes, receptors, or signalling proteins. By constructing and analysing these networks, researchers can uncover the synergistic effects inherent in many TCM prescriptions, where multiple ingredients act on complementary pathways to enhance efficacy or reduce toxicity.

This network-based approach enables a systems-level view of how TCM formulations influence biological pathways, offering insights into the mechanisms of action behind traditional remedies. For example, a single herbal compound might influence inflammatory, immune, and metabolic pathways simultaneously—an effect that would be difficult to isolate without the visual and computational clarity offered by drug-target mapping. Moreover, by integrating drug-target networks with clinical data and disease models, researchers can predict the therapeutic relevance of new herbal combinations, optimise existing formulas for specific syndromes or disease profiles, minimise potential side effects or drug interactions and support personalised TCM treatment strategies.

As AI continues to enhance the construction and analysis of these networks, especially through deep learning and graph-based models, the field moves closer to precision herbal medicine, where treatments are scientifically validated, systematically designed, and tailored to individual biological responses.

Disease-Gene-Drug Networks

Disease-gene-drug networks are powerful tools that integrate genetic, pharmacological, and pathological data to uncover the intricate relationships between disease mechanisms, genetic factors, and therapeutic compounds. Within the context of Traditional Chinese Medicine (TCM), these networks enable researchers to explore how herbal formulations and active compounds influence the underlying genetic causes of disease.

By mapping connections between specific genes associated with diseases and the bioactive ingredients found in TCM, this integrative approach offers several key advantages:

Identification of Genetic Targets: AI-driven network pharmacology helps pinpoint genes central to disease progression. TCM compounds interacting with these genes or their products can be prioritised for further investigation.
Mechanistic Insights: These networks reveal how certain herbs may modulate genetic expression, protein activity, or signalling pathways, providing a scientific rationale for traditional remedies.
Targeted Drug Discovery: By integrating genomics and phytochemistry, researchers can discover or design new TCM-based treatments for specific gene-related pathologies, including chronic and complex diseases such as cancer, autoimmune disorders, and neurological conditions.
Personalised Therapy Development: Disease-gene-drug networks pave the way for precision TCM, allowing herbal prescriptions to be tailored based on a patient’s genetic profile and disease subtype.

Ultimately, by bridging ancient herbal wisdom with modern molecular biology and AI analytics, disease-gene-drug networks elevate TCM to a new level of therapeutic relevance, where treatments are holistic, genetically informed, evidence-based, and clinically adaptive.