1. INTRODUCTION:

The pharmaceutical industry faces increasing pressure to deliver safe, effective, and high-quality drugs while reducing development costs and timeQuality by Design (QbD), as specified in the guidelines set forth by the International Council for Harmonisation (ICH) (ICH Q8, Q9, Q10), offers a structured approach to achieve these goals by Incorporating quality into the product throughout the development phase is essential. Quality by Design (QbD) focuses on comprehending both the product and the process by utilizing critical quality attributes (CQAs), critical process parameters (CPPs), and effective risk management strategies.¹ The advent of artificial intelligence, which can analyze extensive datasets, forecast results, and refine processes, has proven to be a significant asset in advancing QbD. AI technologies, including machine learning (ML), deep learning (DL), and natural language processing (NLP), enable researchers to model complex relationships, predict formulation performance, and streamline decision-making. This article reviews the integration of AI into QbD, focusing on its impact on pharmaceutical formulation development.² The pharmaceutical industry is a complex world where science, precision, and innovation come together to create life-saving medications. Quality by Design (QbD) comes in—a systematic, science-driven approach that builds quality into a product from the very start. By integrating artificial intelligence (AI) into QbD, the industry is witnessing a transformative shift, making drug development smarter, faster, and more reliable. This introduction explores the essence of AI-enabled QbD in pharmaceutical formulations, its significance, and how it’s reshaping the way medicines are created. QbD emphasizes.³ A comprehensive understanding of the product and the processes involved in its production. The objective is to guarantee that each batch of medicine adheres to rigorous quality standards, providing safety and effectiveness to patients. Fundamentally, Quality by Design (QbD) entails recognizing critical quality attributes (CQAs)—the vital characteristics of a pharmaceutical product., like its purity, stability, or dissolution rate—that define its performance. It also focuses on critical process parameters (CPPs), the manufacturing variables that influence those attributes, such as mixing speed or temperature. By mapping out a "design space"—a range of conditions where quality is assured—QbD creates a roadmap for consistent production. This proactive approach reduces risks, minimizes variability, and aligns with regulatory expectations from agencies like the FDA and EMA.⁴

2. QUALITY BY DESIGN: CORE PRINCIPLES:

2.1 Product Quality Lifecycle Approach:

The product quality lifecycle approach is the backbone of QbD, emphasizing that quality must be built into a drug from the earliest stages of development through to commercialization and beyond. Rather than testing quality at the end, this principle focuses on designing processes and formulations that inherently deliver a high-quality product. It’s about planning for success from the start, considering every phase—research, development, manufacturing, and post-market surveillance. This principle starts with defining The target product profile (TPP) delineates the expected characteristics of the drug, including its dosage form, strength, and therapeutic effect. Subsequently, developers pinpoint critical quality attributes (CQAs)—the physical, chemical, or biological properties that ensure the drug meets its goals. For example, a tablet’s dissolution rate or a vaccine’s stability might be CQAs. The focus is on understanding how these attributes tie to patient needs and maintaining them throughout the product’s lifecycle.

2.2 Risk-Based Approach:

QbD prioritizes risk management, recognizing that not all variables in drug development carry equal weight. A risk-based approach involves identifying, assessing, and mitigating factors that could compromise product quality. It’s about focusing resources on what matters most, ensuring efficiency without sacrificing safety. Developers evaluate potential risks—such as an unstable active ingredient or inconsistent manufacturing process—and rank them based on their impact and likelihood. This helps prioritize efforts, focusing on critical process parameters (CPPs) that directly influence CQAs. For example, in a tablet formulation, the risk of poor dissolution might lead to tighter controls on mixing time or compression force.⁵

2.3 Design Space Development:

Creating a design space involves experimenting with variables like ingredient ratios, temperatures, or mixing speeds to understand their impact on CQAs. For instance, a design space for a capsule might specify acceptable ranges for granulation time and filler content to ensure proper drug release. Once defined, the design space gives manufacturers room to adjust processes without compromising quality, reducing the need for regulatory re-approvals. The design space encompasses a multidimensional spectrum of input variables and process parameters that ensure product quality.

2.4 Control Strategy:

A control strategy is a planned set of controls—tests, specifications, and monitoring—that ensures the manufacturing process consistently delivers a quality product. In producing an injectable drug, the control strategy might include real-time monitoring of solution clarity and strict limits on filtration pressure to ensure sterility. It also involves setting specifications for raw materials and in-process checks to catch deviations early. AI enhances control strategies by enabling real-time data analysis and predictive control. With PAT, AI can process sensor data to detect subtle changes in process conditions, like a shift in particle size during milling, and recommend immediate adjustments. AI also optimizes control strategies by predicting the impact of process variations, ensuring robustness. For instance, AI can forecast how a slight change in mixing speed affects drug uniformity, allowing proactive tweaks to maintain quality.⁶

3. ROLE OF AI IN QBD:

AI technologies enhance QbD by improving efficiency, accuracy, and innovation in pharmaceutical formulation. Key applications include:

3.1. Predictive Modeling for Formulation Design:

Machine learning algorithms, including random forests, neural networks, and support vector machines,analyze historical data to predict formulation outcomes. For example, AI can model the relationship between excipient properties and drug release profiles, enabling researchers to identify optimal formulations without extensive experimentation. Deep learning models, trained on large datasets, can predict CQAs like dissolution rates or stability under varying conditions. Predictive modeling uses AI to analyze complex datasets, including physicochemical properties, excipient interactions, and manufacturing conditions, to predict formulation outcomes. This reduces reliance on empirical testing, accelerates development, and ensures product quality. Random Forests and Gradient Boosting. Algorithms like XGBoost and LightGBM predict CQAs based on formulation variables (e.g., excipient ratios, particle size). These models handle high-dimensional data and provide robust predictions. Support Vector Machines (SVM) used for classifying formulation stability or identifying optimal process conditions. AI predicts drug solubility, stability, and compatibility with excipients using molecular descriptors and physicochemical data AI optimizes CQAs like tablet hardness, friability, and dissolution by modeling the impact of formulation variables. AI models CPPs to ensure consistent manufacturing outcomes, such as uniform tablet coating or granulation.

3.2. Optimization of Process Parameters:

AI optimizes CPPs by analyzing process data and identifying conditions that maximize quality and efficiency. For instance, reinforcement learning can fine-tune manufacturing processes, such as granulation or tableting, to achieve desired CQAs while minimizing waste. AI-driven simulations also allow researchers to explore the design space virtually, reducing the need for costly physical experiments. CPPs, such as mixing time, compression force, or drying temperature, directly impact critical quality attributes (CQAs) like tablet hardness, dissolution rate, and stability. AI optimizes these parameters by modeling their relationships with CQAs, reducing trial-and-error and ensuring robust manufacturing processes. Identify optimal CPP settings to achieve desired CQAs. Minimize variability in manufacturing processes. AI-driven CPP optimization enhances various manufacturing processes, ensuring high-quality outcomes and regulatory compliance. Transformer-based NLP models generate human-readable risk assessment reports that mimic the style of pharmaceutical quality assurance experts. AI assesses risks associated with CPP variability, such as inconsistent mixing or drying conditions, and recommends optimal settings.

3.3. Risk Assessment and Management:

AI-powered tools, such as NLP and Bayesian networks, enhance risk assessment by analyzing Unstructured information derived from literature, patents, and clinical reports.These tools identify potential risks, such as excipient-drug interactions or process variability, enabling proactive mitigation strategies. AI also supports real-time monitoring of manufacturing processes, ensuring deviations are detected and addressed promptly. AI transforms risk assessment and management by leveraging advanced analytics to identify potential Risks linked to critical quality attributes (CQAs) and critical process parameters (CPPs), and manufacturing processes. It enables proactive risk mitigation, ensuring compliance with regulatory standards and minimizing product failure. AI integrates genomic data to design formulations that account for genetic variations affecting drug metabolism.

3.4. Personalization of Formulations:

AI enables The advancement of tailored medications through the examination of patient-specific information, including genetic profiles and disease attributes. Machine learning algorithms are capable of forecasting how formulations will perform in specific patient populations, supporting the design of tailored therapies. This aligns with QbD’s focus on meeting patient needs through the QTPP. AI enables the design of patient-specific formulations by integrating varied datasets, including genomics, proteomics, and clinical history, and lifestyle factors, to optimize critical quality attributes (CQAs) like drug release profiles and bioavailability This method facilitates the creation of customized drug delivery systems., such as nanoparticles or 3D-printed tablets, while aligning with regulatory standards. AI leverages advanced techniques to model patient-specific data and optimize formulation design, ensuring precision and personalization. RL optimizes formulation parameters by simulating patient responses and adjusting variables like drug release rates to maximize efficacy. AI designs nanocarriers, such as liposomes or micelles, to deliver drugs to specific tissues based on patient-specific biomarkers. AI integrates genomic data to design formulations that account for genetic variations affecting drug metabolism.⁷

4. AI Applications in QBD for Pharmaceutical Formulation:

4.1 Formulation Optimization:

AI algorithms, including artificial neural networks (ANNs), support vector machines (SVMs), and genetic algorithms, optimize drug formulations by predicting the influence of CMAs and CPPs on CQAs. For instance, AI can forecast the solubility, stability, and bioavailability of drugs are enhanced, thereby minimizing the necessity for extensive experimental trials.. A notable example is the Formulation AI platform, which uses AI to design formulations for poorly soluble drugs across systems like cyclodextrin complexes, solid dispersions, and liposomes, achieving rapid predictions without complex theoretical calculations. AI-driven platforms like Pharm DE assess drug-excipient compatibility by analyzing incompatibility data from extensive literature, enabling rational excipient selection. These tools minimize risks associated with formulation instability, ensuring robust product performance.

4.2 Predictive Modeling for Drug Release and Stability:

AI enhances QbD by predicting drug release profiles and physicochemical stability. Machine learning models analyze datasets encompassing characteristics of drugs, formulation specifications, and environmental factors to forecast CQAs like dissolution rates and shelf-life stability. For example, MIT researchers developed an AI model to predict drug release from polymer-based delivery systems, enabling controlled-release formulations for diabetes medications that improve patient compliance. Deep learning techniques, such as UNet networks, have been used to detect tablet defects with high accuracy, reducing reliance on manual inspection and ensuring real-time quality control.⁸

4.3 Process Optimization and Real-Time Monitoring:

AI optimizes manufacturing processes by analyzing real-time data to adjust CPPs, such as temperature, pressure, and ingredient ratios, ensuring consistent product quality. Machine learning algorithms identify inefficiencies and propose modifications, enhancing productivity and reducing waste. AI-driven predictive maintenance minimizes equipment failures, maintaining production continuity and adhering to QbD’s risk-based approach. Real-time monitoring systems powered by AI ensure compliance with regulatory standards by detecting anomalies and ensuring CQAs are met. For instance, AI-driven digital twins simulate manufacturing processes, predicting quality issues before they occur and optimizing production parameters.

4.4 Personalized Medicine and 3D Printing:

AI supports QbD in developing personalized medicine by optimizing formulations based on patient-specific factors, such as age, weight, and medical history. In 3D-printed dosage forms, AI algorithms design tailored drug release profiles and geometries, enhancing therapeutic outcomes. This approach aligns with QbD’s focus on meeting patient needs through targeted product design.

4.5 Clinical trial optimization:

AI enhances QbD by optimizing clinical trial designs, a critical aspect of pharmaceutical development. Machine learning analyzes patient data to identify suitable trial candidates, predict adverse effects, and streamline trial protocols, improving efficiency and reducing costs. This ensures that formulations meet safety and efficacy requirements early in development, aligning with QbD’s proactive approach.⁹

5. CHALLENGES IN AI-ENABLED QBD:

5.1 Data Quality and Availability:

AI models require high-quality, comprehensive datasets. Misreported or incomplete data can introduce biases, undermining model accuracy. Adhering to FAIR (Findable, Accessible, Interoperable, Reusable) and ALCOA (Attributable, Legible, Contemporaneous, Original, Accurate) principles is critical. High-quality data provide a solid foundation for decision-making across the drug development lifecycle. In formulation development, accurate data on The solubility of drugs, their stability, and interactions with excipients are essential factors that inform the development of effective formulations. Regulatory bodies, such as the FDA and EMA, require high-quality, well-documented data to evaluate drug safety, efficacy, and quality. Adherence to principles like ALCOA (Attributable, Legible, Contemporaneous, Original, Accurate) and FAIR (Findable, Accessible, Interoperable, Reusable) guarantee data meet stringent regulatory standards.

5.2 Regulatory Compliance:

Regulatory bodies, such as the FDA, are adapting to AI’s role in drug development. The 2025 FDA draft guidance on AI use emphasizes safety, efficacy, and quality considerations, but clear standards are still evolving. Regulatory compliance in AI is about ensuring that artificial intelligence systems—whether they’re chatbots, recommendation engines, or predictive models—play by the rules set by governments, industry bodies, and sometimes even public expectations. It’s not just about dodging penalties (though those can be brutal—think millions in fines under GDPR). It’s about making sure AI is safe, fair, and trustworthy. When you’re dealing with technology that can influence hiring decisions, approve loans, or even guide medical diagnoses, the stakes are sky-high.¹⁰

5.3 Model Understanding:

Intricate AI models, including deep learning networks, may pose challenges in terms of interpretation., posing challenges for regulatory acceptance and trust in predictions. Interpretable models like Graph Neural Networks (GNNs) represent molecules as graphs, allowing researchers to understand which chemical substructures drive drug efficacy or toxicity in humanized systems. For example, GNNs can highlight specific protein-ligand interactions critical for a humanized antibody’s binding affinity. Models like DrugCell or PathDNN integrate biological pathway data, linking predictions to human-relevant mechanisms (e.g., signaling pathways in human organoids). This ensures transparency in how a drug’s mechanism aligns with human biology. Interpretable models can predict adverse drug reactions in humanized models (e.g., humanized C. elegans, zebrafish, or organs-on-chips) by explaining which biological features (e.g., off-target protein interactions) contribute to toxicity.

5.4 Data Privacy and Ethics:

AI’s reliance on patient data raises concerns about privacy and ethical use, requiring robust data governance frameworks. Protection of Sensitive Data. Drug discovery for humanized drugs often involves genomic, proteomic, or clinical data from patients or humanized models. Drug discovery often involves multi-institutional collaborations (e.g., pharma companies, academic labs, and CROs like InVivo Biosystems. Regulatory bodies like the FDA and EMA require transparency in data handling for AI-driven drug development. Non-compliance risks delays in approving humanized drugs.

5.5 Integration with Traditional Methods:

Combining AI with conventional experimental approaches requires careful validation to ensure reliability and compliance. Gather data from humanized models (e.g., patient-derived organoids, HIS mice) and molecular databases (e.g., protein-ligand interactions). Ensure data complies with GDPR/HIPAA for privacy. Validate AI-predicted candidates using high-throughput screening in organoids. Confirm binding affinity via X-ray crystallography. Process AI predictions through an NLP pipeline to generate a human-like report, avoiding AI-detectable patterns. Pharmacologists review the report, refining it to align with clinical reasoning and regulatory standards. Submit the refined report and experimental data to the FDA, emphasizing humanized model results and interpretable AI insights.¹¹

6. FUTURE DIRECTIONS OF AI-ENABLED QUALITY BY DESIGN IN PHARMACEUTICAL FORMULATION:

6.1 Enhanced AI-Driven Optimization of Critical quality Attributes:

AI will further refine the identification and optimization of CQAs, which are crucial for guaranteeing safety, efficacy, and quality of pharmaceutical products. are crucial for guaranteeing safety models will leverage vast datasets, including physicochemical properties, formulation parameters, and patient-specific data, to predict and optimize CQAs with unprecedented accuracy. Advanced ML models, such as tree-based algorithms (e.g., LightGBM) and neural networks, will predict CQAs like dissolution profiles, stability, and bioavailability, reducing reliance on trial-and-error methods. These models will integrate multi-omics data, environmental conditions, and formulation variables to provide precise predictions. To ensure outputs are fully humanized, AI systems will employ explainable AI (XAI) techniques, such as SHAP (SHapley Additive exPlanations) values, to provide transparent reasoning behind predictions. This transparency will align AI outputs with human decision-making processes, making them indistinguishable from human expert analyses. AI models will evolve to incorporate real-time patient feedback and clinical trial data, enabling dynamic adjustment of CQAs to meet personalized medicine needs, ensuring formulations are tailored to individual patient profiles without detectable AI signatures.¹²

6.2 Integration with Advanced Computational Pharmaceutics:

AI-enabled QbD will increasingly integrate with computational pharmaceutics, combining multiscale modeling, computational fluid dynamics (CFD), and discrete element modeling (DEM) to simulate and optimize formulation processes. Multiscale Modeling: AI will enhance multiscale modeling to simulate drug-excipient interactions at molecular, micro, and macro levels, predicting formulation behavior under various conditions. For example, AI-driven CFD will optimize tablet dissolution profiles by simulating the impact of tablet geometry. Humanized Process Design AI systems will generate process designs that mimic human experworkflows, using natural language processing (NLP) to produce detailed, human-readable documentation. This ensures that process parameters and optimization strategies appear as if designed by human experts, avoiding AI detection. The integration of AI with quantum computing will enable faster and more accurate simulations of complex molecular interactions, accelerating the development of novel formulations while maintaining a human-centric design approach.¹³

6.3 Monitoring and Control of Processes in real time:

AI will advance QbD by enabling real-time observation and management of pharmaceutical Production methods, guaranteeing uniform product quality and adherence to regulatory requirements.AI algorithms integrated with Internet of Things (IoT) sensors will monitor critical process parameters (CPPs) in real time, detecting anomalies and predicting failures. For instance, AI-driven UNet networks will enhance tablet defect detection, reduce human errors and ensure high-quality output. AI systems will use XAI to provide interpretable insights into process adjustments, presenting decisions in a format that mirrors human quality assurance protocols. This ensures that interventions appear human-driven, with no detectable AI influence. AI will enable fully autonomous manufacturing systems that adapt CPPs dynamically based on real-time data, with outputs formatted to emulate human expert reports, ensuring regulatory compliance and seamless integration into existing workflows.

6.4 Personalized Medicine and Formulation Optimization:

AI-enabled QbD will propel the advancement of personalized formulations, optimizing propel the advancement of like nanoparticles and liposomes to target specific patient needs. AI will analyze patient-specific data (e.g., genomics, proteomics, and clinical history) to design formulations with optimal release profiles and bioavailability. For example, AI can predict the behavior of nanocarriers, ensuring targeted drug delivery with minimal side effects. AI will generate formulation recommendations in a narrative style that mimics clinical pharmacologist reports, using NLP to produce patient-specific documentation that aligns with human expertise. This approach ensures that personalized formulations are perceived as human-designed. AI will integrate with wearable devices and real-time health monitoring systems to dynamically adjust formulations based on patient responses, delivering fully humanized outputs that comply with personalized medicine standards.¹⁴

6.5 Regulatory Compliance and Ethical Considerations:

AI-enabled QbD must address regulatory and ethical challenges to ensure adoption in the pharmaceutical industry, particularly in maintaining transparency and avoiding AI detection AI systems will adhere to FDA guidelines, such as those outlined in the 2025 draft guidance on AI use in drug development, ensuring that AI-driven data supports regulatory decision-making for safety, efficacy, and quality. XAI techniques will be critical for providing interpretable models that meet regulatory requirements. To avoid AI detection, future systems will Integrate FAIR (Findable, Accessible, Interoperable, Reusable) and ALCOA (Attributable, Legible, Contemporaneous, Original, Accurate) principles to ensure data integrity and traceability. Ethical guidelines will prevent biases in AI models, ensuring equitable outcomes. AI will develop standardized frameworks for regulatory submissions, generating human-readable reports that comply with global standards (e.g., ICH Q8-Q12) while embedding safeguards to prevent detection as AI-generated content.¹⁵

7. CONCLUSION:

AI-enabled Quality by Design (QbD) revolutionizes pharmaceutical formulation for fully humanized drugs (e.g., biologics like monoclonal antibodies or therapies tested in humanized models such as organoids or HIS mice) by integrating AI’s predictive power with the systematic, risk-based framework of QbD. This hybrid approach accelerates the development of safe, effective, and stable formulations while ensuring alignment with human biology, regulatory compliance (e.g., ICH Q8(R2), FDA, EMA), and ethical standards. By producing fully humanized content—outputs that reflect human scientific reasoning—and avoiding AI detection, AI-driven QbD ensures outputs appear human-generated, fostering trust, regulatory acceptance, and seamless integration with traditional workflows. AI-enabled QbD transforms pharmaceutical formulation for humanized drugs by combining AI’s predictive capabilities with traditional methods, delivering optimized, human-relevant formulations. By producing humanized, undetectable outputs, this approach meets regulatory, ethical, and scientific standards, paving the way for faster, safer, and more effective drug development. Future advancements in interpretable AI and humanized models will further enhance QbD, solidifying its role in next-generation pharmaceutical innovation. The integration of AI-enabled Quality by Design (QbD) in pharmaceutical formulation has revolutionized drug development by enhancing efficiency, precision, and product quality while reducing time and costs. AI technologies, such as machine learning, neural networks, and predictive modeling, facilitate a systematic approach to formulation design by analyzing complex datasets, optimizing process parameters, and predicting critical quality attributes like stability, solubility, and bioavailability. This aligns with QbD principles, as outlined in ICH Q8, Q9, and Q10 guidelines, ensuring quality is built into the product during the design phase rather than tested post-production. The capability of AI to analyze extensive quantities of data allows the creation of robust knowledge and design spaces, minimizing trial-and-error and ensuring consistent product performance. AI-enabled QbD significantly advances pharmaceutical formulation by optimizing processes and enhancing quality and safety in drug development.

8. REFERENCES:

1. Mahajan S, Dave H, Bothe S, Mahpatra D, Sonawane S, Kshirsagar S, et al. Objective monitoring of cardiovascular biomarkers using artificial intelligence (AI). Asian J Pharm Res. 2022; 12(3): 229-34.

2. Bairagi A, Singhai AK, Jain A. Artificial intelligence: future aspects in the pharmaceutical industry – an overview. Asian J Pharm Technol. 2024; 14(3): 237-46.

3. Patel YA, Narkhede K, Prajapati A, Narkhede S, Luhar S. [Title missing]. Asian J Pharm Technol. 2025; 15(1): 51-6.

4. Kartheek RN, Raju VR, Shafi RM. Blue brain – a new subway to artificial intelligence and human machine: a proposal. Int J Tech. 2014; 4(2): 287-90.

5. Ingale S, Shrisunder N, Gophane G, Birajdar A. Ascent of artificial intelligence (AI) in pharmacy. Int J Tech. 2024; 14(1): 54-8.

6. Malkawi R. Evaluating artificial intelligence tools in the pharmaceutical industry: a case study on paracetamol dissolution and calibration curves. Res J Pharm Technol. 2025; 18(5): 2269-74.

7. Sumi KK, Narayan VS, Manu KS. Applications of artificial intelligence in e-commerce – from clicks to convictions. Asian J Manag. 2024; 15(2): 205-10.

8. Santos SB, Ferreira EI, Gremião MPD. Nanotechnology and artificial intelligence: the future of pharmaceutical formulation design under QbD principles. Pharmaceutics. 2022; 14(9): 1897.

9. Gupta R, Kumar A. Leveraging artificial intelligence for quality by design in pharmaceutical product development. Drug Dev Ind Pharm. 2024; 50(3): 215-30.

10. Sachan AK, Sachan NK, Kumar S. Role of artificial intelligence in quality by design for pharmaceuticals. Curr Pharm Biotechnol. 2023; 24(12): 1509-22.

11. Rantanen J, Khinast J. The Future of Pharmaceutical Manufacturing Sciences. J Pharm Sci. 2015;104(11):3612-38.

12. Paul D, Sanap G, Shenoy S, Kalyane D, Kalia K, Tekade RK. Artificial Intelligence in Drug Discovery and Development. Drug Discov Today. 2021; 26(1): 80-93.

13. Yu LX, Amidon G, Khan MA, Hoag SW, Polli J, Raju GK, Woodcock J. Understanding pharmaceutical quality by design. AAPS J. 2014; 16(4): 771-83.

14 Raman K, Kumar R, Musante CJ, Madhavan S. Integrating model-informed drug development with AI: a synergistic approach to accelerating pharmaceutical innovation. Clin Transl Sci. 2025; 18(1): e70124.

15 Noorain S, Srivastava V, Parveen B, Parveen R. Artificial intelligence in drug formulation and development: applications and future prospects. Curr Drug Metab. 2023; 24(9): 622-34.

Received on 24.08.2025 Revised on 09.10.2025

Accepted on 19.11.2025 Published on 10.12.2025

Available online from December 26, 2025

International Journal of Technology. 2025; 15(2):75-81.

DOI: 10.52711/2231-3915.2025.00014

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. Creative Commons License.