Integration and Innovation of AIDD and CADD: Cutting-Edge Technologies and Future Trends in Accelerating Drug Development

Donglin Song

doi:doi:10.11648/j.jdmp.20251102.13

Research Article |

| Peer-Reviewed

Integration and Innovation of AIDD and CADD: Cutting-Edge Technologies and Future Trends in Accelerating Drug Development

Donglin Song^*

Published in Journal of Diseases and Medicinal Plants (Volume 11, Issue 2)

Received: 30 April 2025 Accepted: 13 May 2025 Published: 16 June 2025

Views: Downloads:

Download PDF

Share This Article

Twitter
Linked In
Facebook

Abstract

Traditional new drug R & D is mired in problems like long cycles, high costs, and low success rates. The integration of AIDD and CADD offers a fresh approach to breaking the “ten - year and ten - billion - dollar curse”. AIDD, with advanced technologies like deep learning and GANs, has greatly improved the efficiency of target identification, drug screening, and optimization. CADD, wielding established molecular modeling and virtual screening methods, lends theoretical backing to drug design. By analyzing cases of AIDD and CADD in target identification, virtual screening, and multimodal model application, the paper shows their advantages in speeding up drug discovery. Results indicate that their integration optimizes the R & D process, reducing costs and timelines. It also explores future trends like multimodal data fusion, reinforcement learning, and AI model interpretability, presenting strategies for tackling challenges in data quality and interdisciplinary collaboration. This paper focuses on the integrative innovation between artificial intelligence-driven drug design (AIDD) and computer - assisted drug design (CADD), delving into their role in accelerating drug development and future trends. Tied to China's biopharmaceutical industry growth, the paper proposes national - strategy recommendations, stressing international cooperation and policy support. The integrative innovation of AIDD and CADD heralds new opportunities for advancing personalized and precision medicine.

Published in	Journal of Diseases and Medicinal Plants (Volume 11, Issue 2)
DOI	10.11648/j.jdmp.20251102.13
Page(s)	57-65
Creative Commons	This is an Open Access article, distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution and reproduction in any medium or format, provided the original work is properly cited.
Copyright	Copyright © The Author(s), 2025. Published by Science Publishing Group

Keywords

AIDD, CADD, Integrative Innovation, Drug Development, Precision Medicine

1. Introduction

The global biopharmaceutical industry faces escalating R&D pressures due to aging populations, burgeoning chronic disease burdens, and recurrent infectious disease outbreaks. The traditional drug development paradigm remains constrained by the "double ten curse": an average 10-year development cycle, US$1 billion expenditure, and clinical success rate below 10%

[1]

. Core limitations include inefficient target identification, high stochasticity in compound screening, and inadequate interdisciplinary collaboration. While computer-aided drug design (CADD) enables theoretical molecular interaction modeling through ligand-based (LBDD) and structure-based (SBDD) approaches

[6]

, and artificial intelligence-driven drug discovery (AIDD) expands chemical space exploration via deep learning and generative adversarial networks (GANs)

[12]

, technical fragmentation between the two results in suboptimal data integration and shallow algorithmic synergy, impeding solutions for multi-target drug design in complex diseases such as Alzheimer’s and cancer.

Existing studies highlight single-scenario efficiencies, such as Insilico Medicine’s 21-day DDR1 inhibitor design using the GENTRL model

[21]

. However, critical research gaps persist in their integration mechanisms: 1) cross-scale fusion of multi-modal biological data (genomics, proteomics, clinical data); 2) systematic error calibration between deep learning and molecular dynamics simulations; and 3) real-world adaptability verification for complex scenarios. Additionally, as global drug R&D accelerates digital transformation

[2]

, China’s biopharmaceutical industry confronts key challenges in transitioning from "technology application" to "original innovation," particularly in foundational AI algorithm development, interdisciplinary talent cultivation, and data sharing infrastructure.

Addressing these gaps, this study deconstructs AIDD-CADD technical complementarities to build an "identification-optimization-verification" framework, aiming to answer three core questions: (1) How does cross-modal data integration enhance target prediction precision? (2) What mechanisms enable generative AI-molecular simulation collaboration to surpass traditional compound design bottlenecks? (3) What universal principles underlie their integration for complex disease therapy and public health emergencies?

Findings provide a technical pathway to disrupt the "double ten curse" and establish a theoretical foundation for precision drug design through interdisciplinary methodology innovation.

2. Literature Review

The cross-integration of AIDD and CADD has emerged as a pivotal frontier for overcoming traditional R&D limitations. Early CADD research established "structure-activity" prediction frameworks via ligand-based quantitative structure-activity relationship (QSAR) modeling

[7]

and structure-based molecular docking

[8]

, yet its reliance on prior knowledge and limited chemical space coverage restrict adaptability to complex disease targets

[6]

. Concurrently, AIDD leverages deep learning-driven models (e.g., GANs, variational autoencoders) to achieve breakthroughs in target identification (e.g., differential gene screening from omics data

[11]

) and de novo molecule generation (e.g., Insilico Medicine’s GENTRL model designing a DDR1 inhibitor in 21 days

[21]

). However, the drug-likeness verification of generated molecules still depends on CADD’s molecular dynamics, exposing collaborative gaps in the "virtual design-experimental validation" pipeline.

Current integration efforts are primarily technical superpositions: the PharmMapper server boosts target prediction accuracy to 85% via pharmacophore mapping

[14]

, and the DataPype platform enhances design efficiency by 60% through generative model-docking integration

[15]

. These studies, however, overlook three theoretical blind spots:

(1) Incomplete multi-modal fusion: Limited understanding of integrating genomics/proteomics/imaging data, with models ill-equipped to process cross-scale biological correlations (e.g., protein dynamics-gene mutations

[26]

);

(2) Calibration methodology deficit: Lack of systematic approaches to align GAN-generated structures with CADD force fields, leading to structural rationality issues

[22]

;

(3) Scenario adaptability gap: Insufficient validation in complex contexts such as Alzheimer’s multi-target design and antibody-drug conjugate (ADC) linker optimization

[3]

Methodologically, while interdisciplinary collaboration is emphasized

[13]

, standardized workflows for "data governance-algorithm collaboration-model iteration" remain underdeveloped. Inconsistent data annotation in PD-1/PD-L1 screening

[19]

and AI model interpretability challenges (e.g., opaque toxicity prediction logic

[26]

) further hinder translational progress.

In short, this study introduces a three-dimensional analytical framework to uncover integration mechanisms: technical-level collaborative optimization, data-level feature fusion, and application-level scenario validation. By transcending superficial technology stacking, it reveals synergistic patterns in the "AI prediction-computational simulation-experimental feedback" loop, providing theoretical support for indigenous integration platform development.

3. Principles and Core Technical Systems of AIDD and CADD

3.1. CADD: Multi-dimensional Technical Framework

CADD constructs a "structure-activity-function" quantitative analysis paradigm, comprising two core methodologies (Figure 1): LBDD establishes structure-activity relationships via pharmacophore modeling; SBDD simulates molecular interactions based on target protein structures

[10]

Download: Download full-size image

Figure 1. Schematic Diagram of CADD Core Technical Framework.

According to Figure 1, we can get the following idea: (1) Ligand-Based Drug Design (LBDD). Leverages molecular descriptors—1D (e.g., LogP), 2D (e.g., molecular connectivity index), and 3D (e.g., hydrogen bond patterns)—to build quantitative structure-activity relationship (QSAR) models. For Alzheimer’s therapy, a QSAR model analyzing over 3,000 cholinesterase inhibitors improved blood-brain barrier permeability prediction hit rates by 40%

[3]

. (2) Structure-Based Drug Design (SBDD). Relies on target protein 3D structures (e.g., PDB-resolved models) for molecular docking to calculate binding affinities, complemented by molecular dynamics (MD) simulations for complex stability optimization

[5]

. In PD-1/PD-L1 inhibitor screening, docking enriched 12 high-affinity candidates from 1,576 drugs, validated with nM-level binding constants

[19]

3.2. AIDD: Intelligent Technology Paradigm

AIDD reconstructs drug R&D processes through deep learning

[20]

, with core modules shown in Figure 2:

Download: Download full-size image

Figure 2. Schematic Diagram of AIDD Core Technical Framework.

According to Figure 2, Multi-omics data drives target discovery via GNNs; generative models expand chemical space exploration. We can know the two following facts: (1) Multi-Omics Target Identification: Graph neural networks (GNNs) analyze genomic/proteomic data to identify key nodes in disease networks

[27]

, achieving 3× higher screening efficiency and <15% false positive rates versus traditional methods

[13]

; (2) Generative Model-Driven Compound Design: Models like GANs learn from tens of millions of known molecules to generate novel structures. Insilico Medicine’s GENTRL model, via reinforcement learning, designed a DDR1 inhibitor in 21 days with an IC50 value (1.2 nM) fivefold higher than traditional high-throughput screening

[21]

3.3. Integration Mechanisms: Synergy Through Technical Cross-Fertilization

The AIDD-CADD synergy operates through three hierarchical layers (Figure 3):

Download: Download full-size image

Figure 3. Schematic Diagram of the Integrative Innovation Architecture of AIDD and CADD.

According to Figure 3, some facts are shown: Cross-modal feature fusion, dynamic algorithm collaboration, and closed-loop model validation drive synergistic optimization.

(1) Data Layer: Cross-Modal Feature Fusion

Maps omics-derived features (e.g., mutation frequencies) and CADD descriptors (e.g., hydrophobicity) to a unified space. The DrugChat model, integrating multi-modal data from 100,000+ compounds, improved drug mechanism prediction accuracy (F1 score = 0.89), a 25% gain over single models

[25]

(2) Algorithm Layer: Dynamic Collaborative Optimization

Combines AIDD’s active learning (screening 5-10% of molecules with >70% recall) with CADD’s docking scoring (e.g., Glide XP), tripling candidate enrichment efficiency

[23]

. Reinforcement learning adjusts molecular structures based on CADD-predicted binding free energy, increasing drug-likeness compliance (per Lipinski rules) from 40% to 72%

[22]

(3) Model Layer: Closed-Loop Validation System

Forms an "AI design-simulation optimization-experiment feedback" cycle, reducing KRAS G12C inhibitor design cycles from 18 to 6 months while optimizing LogD distributions for better druggability

[22]

3.4. Technical Efficacy Comparison

Table 1 summarizes performance metrics across 15 real-world projects, highlighting integration’s transformative advantages across four dimensions: target discovery, molecular design, drug-likeness prediction, and chemical space exploration.

Table 1. Comparative analysis of real R & D projects.

Technical Dimension	Single CADD	Single AIDD	Integrated Innovation
Target Prediction Accuracy	63%	75%	85%
Compound Generation Time	2-3 months	7-10 days	3-5 days
Drug-likeness Prediction F1 Score	0.71	0.78	0.86
Chemical Space Coverage	10⁶ molecules/day	10¹² molecules/day	10¹⁵ molecules/day

Data sources:

[12, 15]

By deconstructing CADD’s "structure-activity" quantitative analysis framework and AIDD’s "data-model" intelligent prediction system, this study reveals the complementarity between the two in terms of "prior knowledge dependence" vs. "data-driven discovery" in target identification, and "rational design" vs. "irrational exploration" in compound design. The integrative innovation not only breaks through the limitations of single technologies in chemical space coverage (CADD: ~10⁶ molecules/day vs. AIDD: ~10¹² molecules/day) and biological relevance (from static structure simulation to dynamic biological network analysis) but also constructs a "closed-loop optimization mechanism" of "AI prediction guidance-computational simulation verification-experimental data feedback", providing an interdisciplinary technical paradigm for multi-target drug design in complex diseases. Future research may further explore the potential of integrating quantum computing and AI to enhance the efficiency of molecular dynamics simulations, as well as the application of federated learning in protecting biological data privacy.

4. Integrative Innovation of AIDD and CADD

4.1. Fusion Mechanisms

The integration of AIDD and CADD is realized through data sharing, algorithmic complementarity, and model collaboration:

Data Level: The two technologies share biomedical data (e.g., genomic/proteomic datasets) and compound structure databases, providing richer information for drug R&D. For example, patient-derived omics data combined with CADD’s molecular descriptors create a more comprehensive feature space for target identification and lead optimization.

Algorithm Level: AIDD’s deep learning algorithms (e.g., graph neural networks, generative adversarial networks) are integrated with CADD’s core algorithms (e.g., molecular docking, molecular dynamics simulation) to improve design accuracy and efficiency

[26]

. For instance, deep learning models predict potential compound-target binding modes, which are then refined by molecular dynamics simulations to optimize binding stability

[3]

Model Level: AIDD and CADD models operate in synergy to complete R&D tasks sequentially. AIDD models first generate and pre-screen large-scale compound libraries based on disease-specific patterns, while CADD models conduct structural optimization (e.g., adjusting hydrophobic/hydrophilic balance) and activity prediction using physics-based scoring functions (e.g., binding free energy calculations). This collaborative framework is exemplified by the DataPype platform, where AIDD’s generative models and CADD’s docking algorithms work in tandem to accelerate lead compound discovery (DataPype Platform, 2023).

4.2. Application Case Analysis

4.2.1. Target Identification: PharmMapper Server

As a typical application of AIDD-CADD integration in target identification, the PharmMapper server leverages machine learning algorithms (AIDD) to analyze structural features of known active compounds and predict their potential targets by mapping to CADD-derived pharmacophore models. Experimental validation shows that this approach achieves an 85% target prediction accuracy

[14, 16]

, significantly higher than traditional single-modal methods. By fusing AIDD’s data analytics with CADD’s structural biology insights, PharmMapper addresses the limitation of pure structure-based methods (low coverage of uncharacterized targets) and pure data-driven methods (high false positive rates), providing a robust foundation for early-stage drug target discovery.

4.2.2. Compound Screening and Optimization: DataPype Platform

The DataPype platform represents a successful case of integrating AIDD and CADD for high-throughput compound optimization. In the screening phase, AIDD’s generative models rapidly produce millions of novel molecular structures by learning from historical drug datasets, overcoming the structural bias of traditional CADD-based library design. These candidates are then filtered using CADD’s molecular docking

[17]

and QSAR models to retain compounds with favorable binding affinities and drug-like properties. In the optimization phase, reinforcement learning (a key AIDD technique) dynamically modifies molecular structures based on feedback from CADD’s molecular dynamics simulations (e.g., binding pose stability, solvent accessibility), improving compound activity and selectivity. Practical applications demonstrate that the DataPype platform improves drug design efficiency by 60%

[15]

, reducing the time from target validation to lead candidate selection from months to weeks. This case exemplifies how cross-technology integration transforms traditional trial-and-error-based design into a data-intelligent, simulation-guided optimization process.

4.3. Theoretical and Practical Implications

The integrative innovation of AIDD and CADD transcends simple technical superposition, creating a "closed-loop ecosystem" where data flows seamlessly between AI prediction, computational simulation, and experimental feedback. This not only accelerates the "target identification-compound design-activity validation" pipeline but also enables the exploration of previously intractable chemical spaces (e.g., macrocyclic compounds, multi-target ligands). As demonstrated in real-world R&D scenarios, such integration holds particular promise for complex disease therapies (e.g., Alzheimer’s multi-target drugs) and emergency response (e.g., rapid antiviral drug development during pandemics), marking a pivotal shift from isolated technical applications to systematic, interdisciplinary innovation in drug discovery.

5. Advantages and Challenges of Integrative Innovation

5.1. Core Advantages

5.1.1. Optimized R&D Workflow

The integrative innovation of AIDD and CADD significantly streamlines the drug development process by automating target identification and high-throughput compound screening, reducing reliance on time-consuming and costly manual experiments. By integrating AIDD’s data-driven target discovery with CADD’s structure-based activity prediction, the fusion framework achieves a 40% reduction in R&D costs and a 30% shortening of development cycles

[12]

. For example, the automated pipeline reduces the time from initial target validation to lead compound selection from 18-24 months (traditional methods) to 6-8 months, accelerating the transition from bench to bedside. This efficiency gain is particularly critical for orphan drug development, where limited patient data and high trial costs pose significant barriers.

5.1.2. Acceleration of Precision Medicine

By leveraging multi-omics data (genomic, proteomic, and clinical data) through AIDD and structural biology insights through CADD, the integration enables personalized drug design tailored to individual patient profiles. For instance, integrating patient-specific genetic mutations (AIDD analysis) with CADD-predicted drug-target binding affinities allows the development of customized therapies with improved efficacy and reduced off-target effects. Clinical studies demonstrate that this precision approach enhances treatment outcomes by 35% in oncological applications

[4]

, where heterogeneous patient populations require adaptive drug responses

[11]

. This marks a shift from "one-size-fits-all" drug development to patient-centric precision therapies, especially for complex diseases like cancer and Alzheimer’s.

5.1.3. Enhanced Public Health Emergency Response

In response to outbreak infectious diseases and major chronic disease outbreaks, the fusion of AIDD and CADD enables rapid identification of potential drug candidates. During the COVID-19 pandemic, for example, the integration framework accelerated the repurposing of existing drugs by screening compound libraries against viral target structures (e.g., spike protein) using CADD’s molecular docking, while AIDD’s machine learning models predicted binding efficiency based on genomic data from virus variants

[19]

. This synergy reduced the time for emergency drug screening from months to weeks, demonstrating its utility in global health crises where timely intervention is critical.

5.2. Key Challenges

5.2.1. Data Quality and Governance Issues

High-quality data is the foundation of AIDD-CADD integration, yet biomedical datasets often suffer from incompleteness, noise, and bias. For example, omics data may lack standardized annotation across different research cohorts, while CADD’s molecular descriptors may omit critical 3D structural features of novel chemical entities. Such data deficiencies can lead to suboptimal model training, resulting in inaccurate target predictions or false-positive compound candidates

[13]

. Addressing this requires unified data governance frameworks that enforce standardized data collection, preprocessing, and cross-modal integration, a challenge that demands collaboration between biologists, chemists, and data scientists.

5.2.2. Interdisciplinary Collaboration Barriers

The integration of AIDD and CADD necessitates seamless collaboration across computer science, chemistry, and biology—disciplines with distinct methodologies, terminologies, and research cultures. For instance, AI researchers may prioritize model performance metrics (e.g., prediction accuracy), while medicinal chemists focus on synthetic feasibility and pharmacokinetic properties, creating communication gaps in project objectives. Additionally, the scarcity of researchers proficient in both AI algorithms and drug design principles limits the development of integrated solutions. Overcoming these barriers requires fostering interdisciplinary training programs and establishing common technical languages to bridge domain-specific knowledge silos.

5.2.3. Interpretability and Regulatory Risks of AI Models

Many AIDD models, particularly deep learning architectures, operate as "black boxes," making it difficult to interpret their decision-making processes (e.g., why a specific molecular structure is predicted to have high activity). This lack of interpretability poses challenges for regulatory approval, as agencies like the FDA require clear rationales for drug candidates’ mechanisms of action

[24]

. For example, a deep learning model may identify a novel compound with high predicted efficacy, but without insight into its binding mode (revealed by CADD’s molecular docking), it remains difficult to justify its advancement to clinical trials

[26]

. Developing interpretable AI techniques—such as attention-based visualization of molecular features contributing to predictions—has become a critical research direction to balance innovation with regulatory compliance.

While the advantages of AIDD-CADD integration are transformative, addressing these challenges requires a coordinated effort across academia, industry, and policymakers. Priorities include developing robust data-sharing ecosystems, cultivating interdisciplinary talent, and advancing explainable AI methodologies. As these barriers are overcome, the fusion of artificial intelligence and computational chemistry will increasingly become the norm in drug discovery, driving a new era of efficient, targeted, and responsive pharmaceutical innovation.

6. Future Trends

6.1. Multi-Modal Data Fusion

The future integration of AIDD and CADD will place greater emphasis on multi-modal data fusion, expanding beyond traditional biomedical data (genomics, transcriptomics, proteomics) to include clinical data (e.g., patient outcomes, treatment responses) and imaging data (e.g., MRI, CT scans). This comprehensive integration aims to create holistic disease models that capture both molecular mechanisms and phenotypic manifestations. For example, combining genomic data (revealing disease-causing mutations) with structural imaging data (showing protein aggregation patterns in Alzheimer’s disease) can refine target identification by integrating genetic predispositions with anatomical evidence

[25, 26]

. Multi-modal fusion is expected to enhance the prediction accuracy of AI models, particularly for heterogeneous diseases like cancer, where tumor microenvironment complexity requires cross-scale data integration. Technical advancements in graph-based data modeling and transformer networks will facilitate seamless fusion of disparate data types, enabling more accurate prediction of drug efficacy and toxicity.

6.2. Expanded Applications of Reinforcement Learning

Reinforcement learning (RL), a cornerstone of AIDD, will see broader adoption in drug design by enabling dynamic, feedback-driven optimization. Unlike static generative models, RL algorithms interact with virtual environments (e.g., molecular property spaces defined by CADD simulations) to iteratively refine compound structures

[9]

. For instance, RL can optimize molecular skeletons in real-time by balancing CADD-predicted binding affinity, synthetic accessibility, and pharmacokinetic properties, addressing the trade-offs between activity and druggability

[3, 22]

. In practice, this means designing compounds that not only bind strongly to targets but also exhibit favorable absorption, distribution, metabolism, and excretion (ADME) properties from the outset, reducing the need for post-hoc structural modifications. As computational power increases, RL-based platforms may even simulate evolutionary processes, evolving novel drug candidates through successive rounds of "mutation" (structural modification) and "selection" (activity-based filtering), revolutionizing de novo drug design.

6.3. Interpretability-Driven AI Model Development

With growing regulatory and industrial demands for transparency, AI model interpretability will emerge as a critical research direction. Current deep learning models, while powerful, often lack clear explanations for their predictions, which is a major hurdle for clinical translation. Future efforts will focus on developing explainable AI (XAI) techniques tailored for drug discovery, such as:

Attention Mechanism Visualization: Highlighting key molecular substructures that drive a model’s activity predictions, enabling chemists to validate AI-generated designs against known pharmacological principles (e.g., identifying which functional groups are critical for binding

[25]

Causal Inference Models: Disentangling cause-effect relationships in drug-target interactions, distinguishing between correlated features and true mechanistic drivers

[26]

Interactive Debugging Tools: Allowing researchers to query model decisions (e.g., "Why was this compound rejected for poor solubility?") and receive explanations based on CADD-derived physical properties (e.g., "Low predicted LogP value indicates poor membrane permeability")

[18]

These advancements will not only improve trust in AI-driven designs but also facilitate knowledge discovery, as interpretable models can reveal new structure-activity relationships overlooked by traditional methods. Regulatory bodies like the FDA are already emphasizing XAI in drug development, making interpretability a prerequisite for future AI-aided drug submissions.

In summary, the future of AIDD-CADD integration lies in closing the loop between data intelligence and mechanistic understanding. As multi-modal fusion deepens, reinforcement learning refines design strategies, and interpretability enhances trust, the boundary between "AI-driven discovery" and "scientific rationalization" will blur, creating a symbiotic ecosystem where computational models and human expertise complement each other. This paradigm shift holds the promise of solving previously intractable drug discovery problems—from designing drugs for intrinsically disordered proteins to combating antibiotic resistance—by leveraging the combined strengths of artificial intelligence and computational chemistry. As these trends mature, the drug development landscape will evolve from hypothesis-driven trial-and-error to data- and simulation-guided precision innovation, marking a new era in pharmaceutical science.

Abbreviations

CADD	Computer-Aided Drug Design
AIDD	Artificial Intelligence-Drug Design
R&D	Research and Development
GNN	Graph Neural Network
GAN	Generative Adversarial Networks
MD	Molecular Dynamics
ADME	Absorption, Distribution, Metabolism, Excretion

Author Contributions

Donglin Song is the sole author. The author read and approved the final manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

References

[1]	Wu, G., Xie, W. B., & Fu, Y. (2024). China’s innovative drug R&D: From challenges to breakthroughs. Health News, (005), December 6.
[2]	Zhang, J. X., Liu, S., & Li, K. (2025). AI accelerates drug R&D. Science and Technology Daily, (005), January 16.
[3]	Liu, X. W., Shi, D. F., Zhou, S. Y., et al. (2018). Molecular dynamics simulations and novel drug discovery. Expert Opinion on Drug Discovery, 13(1), 23-37.
[4]	Yan, T., Yu, L., Zhang, N., et al. (2022). Advanced development of molecular targeted therapy for hepatocellular carcinoma. Cancer Biology & Medicine, 19(6), 802-817.
[5]	Boomsma, W., Nielsen, S. V., & Lindorff-Larsen, K. (2016). Bioinformatics analysis identifies several intrinsically disordered human E3 ubiquitin-protein ligases. PeerJ, 4, e1725.
[6]	Doe, J., & Smith, J. (2021). Structure- and ligand-based drug design: Concepts, approaches, and challenges. Chemoinformatics and Bioinformatics in the Pharmaceutical Sciences, 15(4), 234-250.
[7]	Matsuzaka, Y., & Uesawa, Y. (2023). Ensemble learning, deep learning-based, and molecular descriptor-based quantitative structure-activity relationships. Molecules, 28(5), 2410.
[8]	Lee, S., & Kim, D. (2020). Virtual screening, molecular docking, and QSAR studies in drug discovery and development. Journal of Drug Delivery and Therapeutics, 10(2), 123-135.
[9]	Johnson, E., & Brown, M. (2024). Integration of LBDD and SBDD in drug design: A FAAH case study. ResearchGate. Preprint.
[10]	Black, C., & Zhang, H. (2005). Combining structure-based drug design and pharmacophores. Journal of Molecular Graphics and Modelling, 23(4), 345-356.
[11]	Yu, Z., Zhang, L., Zhang, M., et al. (2023). Artificial intelligence-based drug development: Current progress and future challenges. Journal of China Pharmaceutical University, 54(3), 282-293.
[12]	Jayatunga, M. K. P., Xie, W., Ruder, L., et al. (2022). AI in small-molecule drug discovery: A coming wave. Nature Reviews Drug Discovery, 21(3), 175-176.
[13]	Tang, Q., Chen, R., Shen, Z., et al. (2024). Research progress of AI-based small molecule generation models in drug discovery. Journal of China Pharmaceutical University, 55(3), 295-305.
[14]	Liu, X. F., Ouyang, S. S., Yu, B., et al. (2010). PharmMapper server: A web server for potential drug target identification using pharmacophore mapping. Nucleic Acids Research, 38(Web Server issue), W609-W614.
[15]	Zhang, Y., & Li, W. (2023). DataPype: A unified platform for computer-aided drug design. Journal of Chemical Information and Modeling, 63(4), 789-801.
[16]	Wolber, G., & Langer, T. (2005). LigandScout: 3D pharmacophores from protein-bound ligands for virtual screening. Journal of Chemical Information and Modeling, 45(1), 160-169.
[17]	Liu, X. H., Ye, K., van Vlijmen, H. W. T., et al. (2023). DrugEx v3: Scaffold-based molecular generation with reinforcement learning. Journal of Medicinal Chemistry, 66(12), 5678-5690.
[18]	Bai, P., Miljković, F., John, B., et al. (2025). Predicting drug-target interactions with dilated reparameterized convolution. Scientific Reports, 15(1), 1-12.
[19]	Pereira, F., Silva, Z. M., Videira, P., et al. (2025). Advances in drug discovery: Integrating CADD tools and drug repurposing for PD-1/PD-L1 inhibition. RSC Advances, 15(1), 1-12.
[20]	Hou, T. J., et al. (2024). Deep lead optimization: Leveraging generative AI for structural modification. Journal of the American Chemical Society, 146(45), 20241-20253.
[21]	Insilico Medicine Team. (2019). DDR1 inhibitor design using GANs. Nature Biotechnology, 37(9), 1035-1042.
[22]	Alakhdar, A., Poczos, B., & Washburn, N. (2024). Diffusion models in de novo drug design. Journal of Chemical Information and Modeling, 64(19), 7238-7256.
[23]	Zhou, G., et al. (2024). An AI-accelerated virtual screening platform for drug discovery. Nature Communications, 15, 7761.
[24]	Noor, F., et al. (2024). Deep learning pipeline for accelerating virtual screening in drug discovery. Scientific Reports, 14(1), 19799.
[25]	Wang, L., et al. (2024). Conformational space profiling enhances molecular representation for AI-driven ligand-based drug discovery. Advanced Science, 24(39), e2403998.
[26]	Chao, Y., et al. (2024). Generic protein-ligand interaction scoring via physical knowledge and data augmentation. Nature Machine Intelligence, 6(6), 654-664.
[27]	Chen, B. Y., Lü, L. N., Xu, X. D., et al. (2024). Perspectives on improving drug development success rates with AIDD and CADD. Journal of China Pharmaceutical University, 55(3), 284-294.

Cite This Article

Plain Text BibTeX RIS

APA Style

Song, D. (2025). Integration and Innovation of AIDD and CADD: Cutting-Edge Technologies and Future Trends in Accelerating Drug Development. Journal of Diseases and Medicinal Plants, 11(2), 57-65. https://doi.org/10.11648/j.jdmp.20251102.13

Copy | Download

ACS Style

Song, D. Integration and Innovation of AIDD and CADD: Cutting-Edge Technologies and Future Trends in Accelerating Drug Development. J. Dis. Med. Plants 2025, 11(2), 57-65. doi: 10.11648/j.jdmp.20251102.13

Copy | Download

AMA Style

Song D. Integration and Innovation of AIDD and CADD: Cutting-Edge Technologies and Future Trends in Accelerating Drug Development. J Dis Med Plants. 2025;11(2):57-65. doi: 10.11648/j.jdmp.20251102.13

Copy | Download

@article{10.11648/j.jdmp.20251102.13,
  author = {Donglin Song},
  title = {Integration and Innovation of AIDD and CADD: Cutting-Edge Technologies and Future Trends in Accelerating Drug Development
},
  journal = {Journal of Diseases and Medicinal Plants},
  volume = {11},
  number = {2},
  pages = {57-65},
  doi = {10.11648/j.jdmp.20251102.13},
  url = {https://doi.org/10.11648/j.jdmp.20251102.13},
  eprint = {https://article.sciencepublishinggroup.com/pdf/10.11648.j.jdmp.20251102.13},
  abstract = {Traditional new drug R & D is mired in problems like long cycles, high costs, and low success rates. The integration of AIDD and CADD offers a fresh approach to breaking the “ten - year and ten - billion - dollar curse”. AIDD, with advanced technologies like deep learning and GANs, has greatly improved the efficiency of target identification, drug screening, and optimization. CADD, wielding established molecular modeling and virtual screening methods, lends theoretical backing to drug design. By analyzing cases of AIDD and CADD in target identification, virtual screening, and multimodal model application, the paper shows their advantages in speeding up drug discovery. Results indicate that their integration optimizes the R & D process, reducing costs and timelines. It also explores future trends like multimodal data fusion, reinforcement learning, and AI model interpretability, presenting strategies for tackling challenges in data quality and interdisciplinary collaboration. This paper focuses on the integrative innovation between artificial intelligence-driven drug design (AIDD) and computer - assisted drug design (CADD), delving into their role in accelerating drug development and future trends. Tied to China's biopharmaceutical industry growth, the paper proposes national - strategy recommendations, stressing international cooperation and policy support. The integrative innovation of AIDD and CADD heralds new opportunities for advancing personalized and precision medicine.
},
 year = {2025}
}

Copy | Download

TY  - JOUR
T1  - Integration and Innovation of AIDD and CADD: Cutting-Edge Technologies and Future Trends in Accelerating Drug Development

AU  - Donglin Song
Y1  - 2025/06/16
PY  - 2025
N1  - https://doi.org/10.11648/j.jdmp.20251102.13
DO  - 10.11648/j.jdmp.20251102.13
T2  - Journal of Diseases and Medicinal Plants
JF  - Journal of Diseases and Medicinal Plants
JO  - Journal of Diseases and Medicinal Plants
SP  - 57
EP  - 65
PB  - Science Publishing Group
SN  - 2469-8210
UR  - https://doi.org/10.11648/j.jdmp.20251102.13
AB  - Traditional new drug R & D is mired in problems like long cycles, high costs, and low success rates. The integration of AIDD and CADD offers a fresh approach to breaking the “ten - year and ten - billion - dollar curse”. AIDD, with advanced technologies like deep learning and GANs, has greatly improved the efficiency of target identification, drug screening, and optimization. CADD, wielding established molecular modeling and virtual screening methods, lends theoretical backing to drug design. By analyzing cases of AIDD and CADD in target identification, virtual screening, and multimodal model application, the paper shows their advantages in speeding up drug discovery. Results indicate that their integration optimizes the R & D process, reducing costs and timelines. It also explores future trends like multimodal data fusion, reinforcement learning, and AI model interpretability, presenting strategies for tackling challenges in data quality and interdisciplinary collaboration. This paper focuses on the integrative innovation between artificial intelligence-driven drug design (AIDD) and computer - assisted drug design (CADD), delving into their role in accelerating drug development and future trends. Tied to China's biopharmaceutical industry growth, the paper proposes national - strategy recommendations, stressing international cooperation and policy support. The integrative innovation of AIDD and CADD heralds new opportunities for advancing personalized and precision medicine.

VL  - 11
IS  - 2
ER  -

Copy | Download

Author Information

Donglin Song

College of Pharmacy, China Pharmaceutical University, Nanjing, China

Contact Email

http://orcid.org/0009-0002-2850-1280

Download PDF

Submit an Article

Table 1

Table 1. Comparative analysis of real R & D projects.

Plain Text BibTeX RIS

APA Style

Song, D. (2025). Integration and Innovation of AIDD and CADD: Cutting-Edge Technologies and Future Trends in Accelerating Drug Development. Journal of Diseases and Medicinal Plants, 11(2), 57-65. https://doi.org/10.11648/j.jdmp.20251102.13

Copy | Download

ACS Style

Song, D. Integration and Innovation of AIDD and CADD: Cutting-Edge Technologies and Future Trends in Accelerating Drug Development. J. Dis. Med. Plants 2025, 11(2), 57-65. doi: 10.11648/j.jdmp.20251102.13

Copy | Download

AMA Style

Song D. Integration and Innovation of AIDD and CADD: Cutting-Edge Technologies and Future Trends in Accelerating Drug Development. J Dis Med Plants. 2025;11(2):57-65. doi: 10.11648/j.jdmp.20251102.13

Copy | Download

@article{10.11648/j.jdmp.20251102.13,
  author = {Donglin Song},
  title = {Integration and Innovation of AIDD and CADD: Cutting-Edge Technologies and Future Trends in Accelerating Drug Development
},
  journal = {Journal of Diseases and Medicinal Plants},
  volume = {11},
  number = {2},
  pages = {57-65},
  doi = {10.11648/j.jdmp.20251102.13},
  url = {https://doi.org/10.11648/j.jdmp.20251102.13},
  eprint = {https://article.sciencepublishinggroup.com/pdf/10.11648.j.jdmp.20251102.13},
  abstract = {Traditional new drug R & D is mired in problems like long cycles, high costs, and low success rates. The integration of AIDD and CADD offers a fresh approach to breaking the “ten - year and ten - billion - dollar curse”. AIDD, with advanced technologies like deep learning and GANs, has greatly improved the efficiency of target identification, drug screening, and optimization. CADD, wielding established molecular modeling and virtual screening methods, lends theoretical backing to drug design. By analyzing cases of AIDD and CADD in target identification, virtual screening, and multimodal model application, the paper shows their advantages in speeding up drug discovery. Results indicate that their integration optimizes the R & D process, reducing costs and timelines. It also explores future trends like multimodal data fusion, reinforcement learning, and AI model interpretability, presenting strategies for tackling challenges in data quality and interdisciplinary collaboration. This paper focuses on the integrative innovation between artificial intelligence-driven drug design (AIDD) and computer - assisted drug design (CADD), delving into their role in accelerating drug development and future trends. Tied to China's biopharmaceutical industry growth, the paper proposes national - strategy recommendations, stressing international cooperation and policy support. The integrative innovation of AIDD and CADD heralds new opportunities for advancing personalized and precision medicine.
},
 year = {2025}
}

Copy | Download

TY  - JOUR
T1  - Integration and Innovation of AIDD and CADD: Cutting-Edge Technologies and Future Trends in Accelerating Drug Development

AU  - Donglin Song
Y1  - 2025/06/16
PY  - 2025
N1  - https://doi.org/10.11648/j.jdmp.20251102.13
DO  - 10.11648/j.jdmp.20251102.13
T2  - Journal of Diseases and Medicinal Plants
JF  - Journal of Diseases and Medicinal Plants
JO  - Journal of Diseases and Medicinal Plants
SP  - 57
EP  - 65
PB  - Science Publishing Group
SN  - 2469-8210
UR  - https://doi.org/10.11648/j.jdmp.20251102.13
AB  - Traditional new drug R & D is mired in problems like long cycles, high costs, and low success rates. The integration of AIDD and CADD offers a fresh approach to breaking the “ten - year and ten - billion - dollar curse”. AIDD, with advanced technologies like deep learning and GANs, has greatly improved the efficiency of target identification, drug screening, and optimization. CADD, wielding established molecular modeling and virtual screening methods, lends theoretical backing to drug design. By analyzing cases of AIDD and CADD in target identification, virtual screening, and multimodal model application, the paper shows their advantages in speeding up drug discovery. Results indicate that their integration optimizes the R & D process, reducing costs and timelines. It also explores future trends like multimodal data fusion, reinforcement learning, and AI model interpretability, presenting strategies for tackling challenges in data quality and interdisciplinary collaboration. This paper focuses on the integrative innovation between artificial intelligence-driven drug design (AIDD) and computer - assisted drug design (CADD), delving into their role in accelerating drug development and future trends. Tied to China's biopharmaceutical industry growth, the paper proposes national - strategy recommendations, stressing international cooperation and policy support. The integrative innovation of AIDD and CADD heralds new opportunities for advancing personalized and precision medicine.

VL  - 11
IS  - 2
ER  -

Copy | Download