ARTIFICIAL INTELLIGENCE IN DRUG DEVELOPMENT

ARTIFICIAL INTELLIGENCE IN DRUG DEVELOPMENT

Pranshu Patel, Shiv Hardenia * and Dinesh Kumar Jain

IPS Academy College of Pharmacy, Knowledge Village, Rajendra Nagar, A.B. Road, Indore, Madhya Pradesh, India.

ABSTRACT: Artificial intelligence (AI) is rapidly transforming the drug development process, offering innovative solutions to streamline and accelerate the discovery of new therapeutics. Traditionally, drug development has been a time-consuming and costly endeavor, often taking over a decade and billions of dollars to bring a drug to market. AI, with its ability to analyze vast amounts of complex data, has the potential to significantly reduce both time and costs while improving the success rate of new drugs. AI techniques, including machine learning (ML), deep learning (DL), and natural language processing (NLP), are being integrated at various stages of the drug development pipeline. In the early stages, AI is used to analyze large datasets, such as chemical, genomic, and proteomic data, to identify potential therapeutic targets and biomarkers. Machine learning algorithms can also predict a compound’s biological activity, optimizing lead compounds and enhancing virtual screening compared to traditional methods. Advanced AI methods, such as generative models and reinforcement learning, are being applied to design new molecules with desired properties, improving drug efficacy and enabling drug repurposing. AI also aids in predicting pharmacokinetics, toxicity, and drug-target interactions, improving safety profiles and reducing clinical trial failures. In clinical development, AI helps personalize treatment plans by evaluating patient data, optimizing clinical trial procedures, patient recruitment, and trial design. Despite challenges like data quality and regulatory concerns, AI holds the potential to revolutionize drug development, enabling faster, more effective treatments for patients.

Keywords: Artificial intelligence, Drug discovery, Drug development, Drug repurposing, Polypharmacology

INTRODUCTION: Drug development involves the comprehensive process of bringing a new drug molecule into clinical use. In its broadest sense, it covers all stages, from fundamental research focused on identifying suitable molecular targets to large-scale Phase III clinical trials that facilitate the drug's commercial launch, along with post-market pharmacovigilance and studies on drug repurposing ¹.

The drug development process includes identifying and rigorously testing chemical entities with potential as therapeutic agents, and it is time-consuming as well as costly ². It is expensive because high cost associated with R&D and clinical trials. On an average, it takes approximately 12 to 15 years for bringing a new drug molecule to the market for patient treatment.

The rate of drug production in the pharmaceutical industry is still on the decline. This tendency can be attributed to a number of factors, such as the existing saturation of the market, the challenges of introducing novel chemical matter through a convoluted approvals procedure, and the willingness-to-pay in both developed and developing countries.

The inherent difficulty of transferring the drug development process from basic science to early clinical trials will be the main topic of discussion here. More information on a variety of topics related to the field is available to scientists now than ever before, greatly exceeding their capacity to appropriately interpret and incorporate into their own workflows and research goals ³. To overcome these challenges, researchers globally have increasingly adopted computational methods such as virtual screening (VS) and molecular docking, commonly known as traditional approaches. Nevertheless, these techniques also face challenges, including problems with accuracy and efficiency ⁴. As a result, there is a growing interest in adopting innovative techniques that can effectively address the challenges faced by traditional computational methods. Artificial intelligence (AI), including deep learning (DL) and machine learning (ML) algorithms, has emerged as a potential solution to overcome the obstacles in drug design and discovery ⁵.

AI, Machine Learning, and Natural Language Process (NLP): AI is a branch of computer science that uses math, logic, and cognitive science to simulate human behavior. The integration of AI into the healthcare industry can be summarized as the application of techniques that allow computers to imitate human behavior. It includes machine learning (ML) and natural language processing (NLP). Further subset of ML is deep learning (DL). ML uses statistical methods to enable systems to learn independently, without explicit programming. It is divided into three types: supervised, unsupervised, and reinforcement learning. Supervised learning builds predictive models using input and output data, with methods like classification (for disease diagnosis) and regression (for predicting drug efficacy and ADMET properties) ⁶. Unsupervised learning includes feature-finding and clustering techniques that categorize and analyze data exclusively from input data ⁷. Reinforcement learning is a type of machine learning used in complex environments to optimize decision-making through trial and error, based on feedback. While similar to supervised learning in its goal of improving decisions, it differs in its learning approach ⁸.

Deep learning (DL) is a subset of machine learning that excels in generalization, feature extraction, and end-to-end learning. In DL, output is derived directly from input during training, with the model refined through iterations based on prediction errors. A deep neural network (DNN) is a feed-forward network with interconnected input, hidden, and output layers, using multiple neurons to model complex, nonlinear relationships and learn features from input data ⁹.

FIG. 1: CLASSIFICATION OF ARTIFICIAL INTELLIGENCE

Second class of AI is natural language processing (NLP) techniques that enhance and augment organized medical data by extracting information from unstructured sources like clinical notes and medical journals. The goal of NLP processes is to convert texts into structured data that is machine-readable so that ML approaches can analyze them ¹⁰.

History of AI: The idea of using computers to mimic human behavior and intelligence was initially proposed by Turing six years prior, but McCarthy first defined artificial intelligence (AI) in 1956 as "the science and engineering of making intelligent machines ¹¹." John Mc Carthy, an American computer scientist was called the father of Artificial Intelligence.

TABLE 1: A BRIEF HISTORY OF AI DEVELOPMENT

Drug Development Process: The FDA defines the drug development process as consisting of four stages: (i) drug discovery, which involves identifying new therapeutic agents by understanding disease mechanisms; (ii) preclinical research, where laboratory and animal testing are conducted to assess the safety of new drug targets; (iii) clinical research, which includes several phases of human clinical trials to evaluate the drug's safety and effectiveness; and (iv) post-marketing research, focusing on comparative effectiveness studies and pharmacosurveillance to monitor the drug's performance after it reaches the market.

The drug development process begins by collecting data from computational modeling, high-throughput screening, and research. Researchers use inductive and deductive reasoning to identify potential drug compounds. Automation of steps like virtual screening and in-silico synthesis improves efficiency and accuracy, predicting toxicity and effectiveness without biochemical testing ⁶. The drug development process begins by identifying a hit compound for optimization. High-throughput screening selects hits based on statistical analysis or inhibition thresholds, while fragment-based screening adjusts activity by molecular size to ensure potency and suitability ¹⁹. Once a lead compound is identified, its efficacy and safety are evaluated in animal models. Further preclinical research assesses pharmacokinetics and toxicity. If results are promising, an Investigational New Drug (IND) application is submitted to the FDA for approval to begin human clinical trials. Clinical trials proceed in four phases: Phase 1 focuses on safety and dosage; Phase 2 evaluates efficacy and side effects; Phase 3 confirms efficacy in a larger population; and Phase 4, or post-marketing surveillance, tracks long-term effects. Before Phase 4, a New Drug Application (NDA) is submitted to the FDA for marketing approval.

AI in Drug Discovery and Drug Development: One of the biggest challenges in drug discovery is identifying effective and bioactive drug candidates from the vast chemical space of approximately 10^60 potential molecules, with 90% of candidates failing to advance past Phase II clinical trials or gain regulatory approval. However, these challenges can be mitigated by integrating AI-driven tools throughout the drug development process. AI helps design small molecules, optimize dosages, predict bioactive agents, analyze protein interactions and folding, perform virtual screening, conduct QSAR modeling, facilitate drug repurposing, and assess toxicity, bioactivity, and the mode of action of drug substances ⁵. AI enhances outcomes in areas where traditional methods struggle, particularly when assessing complex properties like drug-likeness, which rely on the interplay of multiple factors.

Optimizing these compounds is a complex process, often supported by in-silico prediction techniques. Machine learning algorithms like Bayesian learning, Random Forests (RF), and support vector machines (SVM) are commonly employed to enhance the optimization process ²⁰.

Further advancements in machine learning have improved drug discovery, especially through QSAR-based ligand-based virtual screening (LBVS). SVMs, Bayesian algorithms (including naïve Bayes), and Random Forests (RF) are key machine learning techniques in QSAR modeling, helping to identify potential therapeutic candidates by analyzing complex biological data and molecular interactions. These methods are advancing AI-driven tools in drug discovery ³. While these models are excellent for predicting basic properties like logP and solubility, they struggle with more complex aspects like side effects and efficacy. Issues such as insufficient experimental validation, data errors, and small training sets often lead to overfitting, limited chemical space, and inaccurate predictions. Moreover, QSAR's assumption that structurally similar compounds will have similar activities is sometimes flawed, highlighting the need for more comprehensive models ²¹. Combinatorial library screening includes both virtual and experimental methods. Virtual screening uses computational techniques like molecular docking, pharmacophore mapping, and QSAR to predict drug-target interactions. However, it has limitations, such as not fully replacing experimental screening and the challenge of synthesizing compounds. On the other hand, experimental methods like high-throughput screening (HTS) directly assess the biological activity of large numbers of molecules, providing more reliable results. Since, the 1990s, HTS has revolutionized drug discovery by enabling fast screening of vast chemical libraries with minimal resources, often using robotics. This has reduced testing costs and contributed to big data, providing valuable insights into drug responses to specific targets ²².

AI in Virtual Screening: The advancement of virtual screening methods has slowed, limiting improvements in prediction accuracy. While virtual screening can identify marginally active ligands, optimizing them is challenging ³. For initial compound selection, Richards and colleagues' new study demonstrates the efficacy of utilizing coarse-grained models. Virtual screening is broadly distinguish into two category: ligand based virtual screening (LBVS) and structured based virtual screening (SBVS).

Structure Based Virtual Screening: Structure-based virtual screening (SBVS) is a computational approach in drug development that identifies potential therapeutic candidates by predicting how small molecules (ligands) interact with target proteins (receptors) ²³.  It selects compounds likely to bind to the protein and exhibit biological activity, based on the 3D structure of the target protein, which is often determined through techniques like X-ray crystallography, NMR spectroscopy, or computational methods such as AlphaFold ^{24, 25}.

Ligand Based Virtual Screening: Ligand-based virtual screening (LBVS) relies on the structural and physicochemical properties of known active and inactive compounds, based on the principle that similar structures often have similar biological activities. AI-powered ligand-based methods are now the most widely used in virtual screening ^{3, 26}. In LBVS, the similarity between compounds in a library and known active compounds is assessed using molecular descriptors, which can be 1D and 2D (providing chemical properties and topological characteristics) or 3D (describing molecular fields, shape, volume, and pharmacophores). LBVS is particularly useful when the 3D structure of the target protein is unavailable, such as for G-protein-coupled receptors (GPCRs) or proteins in their apo form ^{23, 26, 27}.

AI in Drug Design: Predicting 3D structure of target protein - Most pharmacological targets are proteins, and their structures are crucial for biological functions like cell signaling and enzymatic activity. While techniques such as NMR spectroscopy, cryo-EM, and X-ray crystallography can determine protein structures, they are costly and time-consuming, with only around 100,000 structures identified so far. To address this gap, new methods for determining protein structures are needed ²⁸.

AI-based methods have become increasingly popular for predicting protein structures, driven by the availability of large protein datasets ²⁹. One early approach by Qian et al. (1988) used a neural network to predict protein secondary structures with a success rate of 64.3%. The complexity of protein structure prediction stems from the vast conformational space, often breaking down the problem into components like secondary (2°) structure, solvent-accessible surface area, backbone torsion angles and other one-dimensional (1D) structural properties ³⁰. Structure-based drug design (SBDD) relies on understanding protein structure to identify therapeutic compounds, but experimental methods to determine 3D structures are costly and time-consuming ³¹.

The most advanced tool for predicting protein 3D structures is DeepMind's AlphaFold, a neural network-based technology that achieves accuracy comparable to experimental methods ³². The details of the algorithm and architecture of AlphaFold are outlined in the work by Senior et al ³³. To improve protein structure prediction, AlphaFold employs sophisticated neural networks that combine geometric, physical, and evolutionary constraints ^{32, 34}. De novo structure prediction aids structural genomics by filling in missing protein segments, identifying novel folds, and modeling proteins difficult to analyze with X-ray crystallography or NMR, providing early functional insights before experimental determination ³⁵.

Predicting Drug Protein Interaction: In drug development, understanding how molecules interact with proteins in cellular networks is key to creating effective treatments ³⁶. Drug-protein interaction (DPI) prediction plays a crucial role in drug development, with two main approaches: machine learning and physics-based techniques. Physics-based methods, such as molecular docking, model interactions at the atomic level using energy functions, but they are complex due to protein size and folding ³⁷. Protein folding, the process of predicting interactions from a linear 1D sequence to a 3D protein structure. Traditional 1D sequence methods struggle to capture the necessary 3D structural details for accurate DPI predictions ³⁸. Various methods have been developed to predict DPI, including docking simulations, literature mining, and techniques that integrate chemical, 3D structural and genomic data. Structural bioinformatics and protein cleavage site prediction are also used to improve prediction accuracy which is based on Chou’s distorted key theory.³⁹ These methods aim to enhance the understanding of how drugs interact with proteins like enzymes, ion channels, and GPCRs. High-throughput techniques examining the genome, transcriptome, and proteome are further advancing our knowledge of these interactions ⁴⁰. Key drug-target protein interaction data has been collected from databases such as KEGG BRITE, BRENDA, SuperTarget, and DrugBank, which track interactions with enzymes, ion channels, GPCRs, and nuclear receptors ⁴¹.

FIG. 2: AI IN DRUG DISCOVERY

AI in De Novo Drug Design: The term "de novo," meaning "from scratch", De novo drug design (DNDD) creates completely new chemical molecule, without using existing templates. It aims to explore uncharted chemical space and generate new compounds that meet specific biological requirements. Recently, deep reinforcement learning (DRL) has been integrated into DNDD, combining reinforcement learning and neural networks to enhance the design process ⁴². Unlike methods like virtual screening, which optimize pre-existing molecules, de novo design creates entirely new chemical entities that do not exist in nature ⁴³. Key challenges in DNDD include navigating the vast chemical space and extracting desired properties. AI techniques, such as variational autoencoders (VAEs), generative adversarial networks (GANs), and recurrent neural networks (RNNs) with long short-term memory (LSTM) cells, have been employed to guide molecule design using molecular graphs or SMILES notation ⁴⁴. In DRL-based de novo drug design, an RNN is first trained on bioactive chemicals, such as those in the ChEMBL database, to predict molecular sequences. The model is then used to generate new sequences, while a drug-design agent learns from reinforcement learning to optimize the drug design process. The agent's policy is refined over time to maximize the expected return, making this approach highly effective in generating novel, bioactive compounds ^{41, 45}.

AI in Drug Screening: Advancements in AI, big data, and deep learning are transforming drug discovery by speeding up target identification, optimizing clinical trial success, and improving drug design efficiency. AI integration with high-throughput screening allows faster data processing, boosting the accuracy and success rate of drug development. Key collaborations, such as the UK’s 100,000 Genomes Project and DeepMind’s work with the Royal Free London NHS, showcase AI’s growing impact on diagnostics, drug design, and treatment of rare diseases ^{46, 47}.

Prediction of the Physicochemical Properties: Physicochemical properties like partition coefficients (logP), solubility (logS), melting and boiling points, vapor pressure, and bioconcentration factor (BCF) are crucial for understanding how chemicals behave in both the environment and biological systems. These properties influence exposure, toxicological risks, bioavailability, absorption, and persistence in the body and environment ⁴⁸. In medicinal chemistry, algorithms convert a molecule's IUPAC name into SMILES notation, representing its structure with ASCII strings, which can be visualized as 2D or 3D models. Using pattern-matching algorithms and databases like PubChem, known chemical structures can be identified, and new molecules can be designed ⁴⁹.

Lipophilicity is a key factor in drug development, affecting pharmacokinetics, membrane permeability, and ADME behavior. Methods like logP, logD, and IAM algorithms estimate lipophilicity, while computational techniques such as group contribution methods, quantum chemistry, and QSAR models are also used. The average of predictions from models like XLOGP3, WLOGP, and iLOGP is often used to refine lipophilicity estimates ⁸. Machine learning approaches in ADME research are helping to predict the pharmacokinetic properties of new drug candidates, optimizing molecules by forecasting bioavailability ⁵⁰. Prediction of bioactivity: AI has significantly advanced drug development, particularly in drug bioactivity prediction ³⁴. High-throughput screening (HTS) technologies, including mass spectrometry-based proteomics, genomics, and advanced chromatographic techniques, have accelerated drug discovery from natural products and enhanced biomedical insights ⁵¹. The use of quantitative high-throughput screening (qHTS), which tests compounds at varying doses, has driven the development of activity-based approaches in drug discovery, generating valuable data for chemical activity forecasting. Over the past 15 years, more than half a million compounds have been screened, producing activity profiles that support biological activity-based modeling (BABM). Collections like LOPAC and the NCATS Pharmaceutical Collection (NPC) provide extensive datasets for training machine learning models ⁵².

In AI-driven drug discovery, molecules are represented by SMILES notation to form molecular graphs. Feature vectors are created based on bond attributes (such as bond type, stereochemistry) and atomic properties (such as bond count, atomic number). After refinement through message-passing algorithms, these feature vectors are fed into a neural network to predict the likelihood of a molecule having antibacterial properties ²⁸.

Predicting toxicity: Recent online tools such as REACHacross2 and the Chemical In Vitro–In Vivo Profiling portal (CIIPro)enable automated extraction of data from databases like REACH and PubChem. Tools like Chembench and Charming support the development and sharing of curated toxicity data and QSAR models ⁵³. The U.S. EPA's distributed structure searchable toxicity (DSSTox) Database Network and the Vitic Toxicity Database, supported by pharmaceutical and chemical companies through Health and Environmental Science Institute (HESI) and managed by Lhasa Limited, store extensive toxicology data that can be re-analyzed using various techniques ⁵⁴. AI techniques, including DeepTox and PrOCTOR, are essential for predictive toxicology, helping assess toxicity risks in clinical trials and forecast the safety of new drugs. For example, Robledo-Cadena et al. (2020) was used PrOCTORto predict how NSAIDs affect the efficacy of cancer drugs like doxorubicin and paclitaxel, while AI models helped explore new therapeutic indications for compounds targeting Type 2 diabetes and Parkinson's disease ⁵.

AI in Drug Repurposing: The high cost of drug development has led to a shift from the traditional "one disease, one target, one drug" model, making drug repurposing a more feasible strategy. By integrating genomic, proteomic, and metabolomic data, researchers can identify new targets for existing drugs ⁶³.

In-silico methods using transcriptome data from various biological systems, along with deep learning, can predict a drug's potential for repurposing ⁶. Drug repurposing often begins with unexpected findings, such as axitinib, initially developed for advanced renal cell carcinoma, showing activity against acute lymphoblastic leukemia (ALL) and chronic myeloid leukemia (CML) cells. Studies revealed axitinib binds strongly to the T315I-mutated BCR-ABL1protein, common in these cancers ⁵⁵. AI techniques like PREDICT, SLAMS, NetLapRLS, and DTINet combine data from multiple sources to identify potential drug repurposing opportunities ⁶.

AI in Polypharmacology: Polypharmacology, defined by the National Library of Medicine as the design or use of pharmaceutical agents that target multiple disease pathways, is integral to the development of multi-target treatments like CAR-T cells, bispecific monoclonal antibodies (BsMAb) and PROTACs ⁵⁶. Drug repurposing offers several advantages, including lower clinical failure rates, faster discovery, and reduced development costs by leveraging established safety profiles, thus avoiding redundant safety testing ⁵⁵. Various databases, such as ChemSpider, ZINC, PubChem, KEGG, ChEMBL, and DrugBank, provide extensive information on molecular pathways, structures, and biological activities, all of which can be explored by AI to develop polypharmacological agents. A prime example is the DeepDDI platform, which predicts drug-drug interactions and identifies alternative medications with minimal side effects for clinical use ^{6, 55}.

TABLE 2: LIST OF AI BASED SOFTWARE AND TOOLS FOR DRUG DISCOVERY

AI in Clinical Trial: Clinical trial failures, which drive up costs and delays in drug development, are largely due to inefficient patient recruitment and monitoring.

AI and machine learning (ML) are being increasingly used to improve patient selection, predict treatment success, and reduce variability, thus enhancing trial efficiency and success rates ⁶¹. AI helps identify patients based on biomarkers and supports drug discovery in areas such as drug targets, small molecules, biologics, vaccines, and repurposed drugs.

For example, IBM Watson uses AI to match patients with trials and predict outcomes, improving trial design and reducing costs. Ongoing AI advancements are expected to further optimize clinical trials and aid in personalized medicine development ^{61, 62}.

FIG. 3: AI IN DRUG DISCOVERY AND DRUG DEVELOPMENT PROCESS

AI in Pharmaceutical Manufacturing: Pharmaceutical formulation, which involves combining chemical compounds to create effective drugs, is a critical step in preclinical studies. AI is increasingly used to predict the properties of both solid and non-solid dosage forms, with a particular focus on solid forms like tablets. AI models help predict characteristics such as tensile strength, disintegration time, and brittleness based on factors like active pharmaceutical ingredient (API) properties, excipient types, and coating materials ⁶³. AI has enhanced the efficiency, accuracy, and cost-effectiveness of formulation development by analyzing experimental data to identify correlations between formulations, operational parameters, and quality attributes. This accelerates the optimization process, although if a formulation doesn't meet specific criteria (such as API loading, stability), it may need refinement, requiring multiple steps ^{64, 65}. Machine learning (ML), a subset of AI, is especially useful in predicting outcomes like drug stability by analyzing past data. With improved algorithms, faster computing, and more accessible tools, ML has become integral to pharmaceutical development, streamlining formulation processes and improving results ⁶⁵.

CONCLUSION: Drug development is being revolutionized by AI, which increases cost-effectiveness, efficiency, and precision. It speeds up drug development, makes personalized treatment possible, improves predictive analytics, and streamlines clinical trials. AI has many advantages, but there are also problems, such as data privacy and legal concerns. When AI and human knowledge work together, its full potential will be achieved, offering patients quicker and more efficient treatments.

ACKNOWLEDGMENT: We are expressing our sincere gratitude and respect towards IPS Academy College of Pharmacy, Indore, and their faculty for their time and effort in reviewing this article. Their valuable feedback, constructive criticism, and insightful suggestions have greatly contributed to improving the quality of the content. I deeply appreciate their expertise and thoughtful input during the review process.

CONFLICTS OF INTEREST: The authors have no conflicts of interest.

REFERENCES:

1. Chen Z, Liu X, Hogan W, Shenkman E and Bian J: Applications of artificial intelligence in drug development using real-world data. Drug Discovery Today 2021; 26(5): 1256-64.
2. Tiwari PC, Pal R, Chaudhary MJ and Nath R: Artificial intelligence revolutionizing drug development: Exploring opportunities and challenges. Drug Development Research 2023; 84(8): 1652-63.
3. Yang X, Wang Y, Byrne R, Schneider G and Yang S: Concepts of artificial intelligence for computer-assisted drug discovery. Chemical Reviews 2019; 119(18): 10520-94.
4. Stanzione F, Giangreco I and Cole JC: Use of molecular docking computational tools in drug discovery. Progress in Medicinal Chemistry 2021; 60: 273-343.
5. Gupta R, Srivastava D, Sahu M, Tiwari S, Ambasta RK and Kumar P: Artificial intelligence to deep learning: machine intelligence approach for drug discovery. Molecular Diversity 2021; 25: 1315-60.
6. Mak KK and Pichika MR: Artificial intelligence in drug development: present status and future prospects. Drug Discovery Today 2019; 24(3): 773-80.
7. Gangwal A, Ansari A, Ahmad I, Azad AK and Sulaiman WM: Current strategies to address data scarcity in artificial intelligence-based drug discovery: A comprehensive review. Computers in Biology and Medicine 2024; 179: 108734.
8. Selvaraj C, Chandra I and Singh SK: Artificial intelligence and machine learning approaches for drug design: challenges and opportunities for the pharmaceutical industries. Molecular Diversity 2021; 1-21.
9. Lu M, Yin J, Zhu Q, Lin G, Mou M, Liu F, Pan Z, You N, Lian X, Li F and Zhang H: Artificial intelligence in pharmaceutical sciences. Engineering 2023; 27: 37-69.
10. López-Úbeda P, Martín-Noguerol T, Aneiros-Fernández J and Luna A: Natural language processing in pathology: current trends and future insights. The American Journal of Pathology 2022; 192(11): 1486-95.
11. Gallego V, Naveiro R, Roca C, Ríos Insua D and Campillo NE: AI in drug development: a multidisciplinary perspective. Molecular Diversity. Springer Science and Business Media LLC 2021; 1461–79.
12. Khan FH, Pasha MA and Masud S: Advancements in microprocessor architecture for ubiquitous AI An overview on history, evolution, and upcoming challenges in AI implementation. Micromachines 2021; 12(6): 665.
13. Badiru AB and Asaolu O: Handbook of Mathematical and Digital Engineering Foundations for Artificial Intelligence: A Systems Methodology CRC Press 2023; 416.
14. Xu Y, Liu X, Cao X, Huang C, Liu E, Qian S, Liu X, Wu Y, Dong F, Qiu CW and Qiu J: Artificial intelligence: A powerful paradigm for scientific research. The Innovation 2021; 2(4): 1-20.
15. Kaur GK, Malik N, Chauhan S and Kaur M: Artificial Intelligence Overview: Architecture, Applications and Challenges. Reshaping Intelligent Business and Industry: Convergence of AI and IoT at the Cutting Edge 2024; 15: 1-8.
16. Niazi SK: The coming of age of AI/ML in drug discovery, development, clinical testing, and manufacturing: The FDA Perspectives. Drug Design, Development and Therapy 2023; 2691-725.
17. Ding H, Wu J, Zhao W, Matinlinna JP, Burrow MF and Tsoi JK: Artificial intelligence in dentistry A review. Frontiers in Dental Medicine 2023; 4: 1-13.
18. Kaul V, Enslin S and Gross SA: History of artificial intelligence in medicine. Gastrointestinal Endoscopy 2020; 92(4): 807-12.
19. Kumar N and Acharya V: Machine intelligence-driven framework for optimized hit selection in virtual screening. Journal of Cheminformatics 2022; 14(1): 1-21.
20. Wang L, Song Y, Wang H, Zhang X, Wang M, He J, Li S, Zhang L, Li K and Cao L: Advances of artificial intelligence in anti-cancer drug design: a review of the past decade. Pharmaceuticals 2023; 16(2): 253.
21. Nasnodkar S, Cinar B and Ness S: Artificial intelligence in toxicology and pharmacology. Journal of Engineering Research and Reports 2023; 25(7): 192-206.
22. Haider R: Drug Discovery in the 21st Century. Progress Annals: Journal of Progressive Research 2023; 1(1): 33-50.
23. Bhunia SS, Saxena M and Saxena AK: Ligand-and structure-based virtual screening in drug discovery. In Biophysical and Computational Tools in Drug Discovery 2021; 281-339.
24. Kumar H and Kim P: Artificial intelligence in fusion protein three‐dimensional structure prediction: Review and perspective. Clinical and Translational Medicine 2024; 14(8): 1789.
25. Oliveira TA, Silva MP, Maia EH, Silva AM and Taranto AG: Virtual screening algorithms in drug discovery: A review focused on machine and deep learning methods. Drugs and Drug Candidates 2023; 2(2): 311-34.
26. Vázquez J, López M, Gibert E, Herrero E and Luque FJ: Merging ligand-based and structure-based methods in drug discovery: An overview of combined virtual screening approaches. Molecules 2020; 25(20): 4723.
27. Wu H, Liu J, Zhang R, Lu Y, Cui G, Cui Z and Ding Y: A review of deep learning methods for ligand based drug virtual screening. Fundamental Research 2024; 715-737.
28. Chen W, Liu X, Zhang S and Chen S: Artificial intelligence for drug discovery: Resources, methods, and applications. Molecular Therapy-Nucleic Acids 2023; 31: 691-702.
29. Tripathi N, Goshisht MK, Sahu SK and Arora C: Applications of artificial intelligence to drug design and discovery in the big data era: a comprehensive review. Molecular Diversity 2021; 25(3): 1643-64.
30. Elofsson A: Progress at protein structure prediction, as seen in CASP15. Current Opinion in Structural Biology 2023; 80: 102594.
31. Vemula D, Jayasurya P, Sushmitha V, Kumar YN and Bhandari V: CADD, AI and ML in drug discovery: A comprehensive review. European Journal of Pharmaceutical Sciences 2023; 181: 106324.
32. Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, Tunyasuvunakool K, Bates R, Žídek A, Potapenko A and Bridgland A: Highly accurate protein structure prediction with AlphaFold. Nature 2021; 596(7873): 583-9.
33. Stahl K, Graziadei A, Dau T, Brock O and Rappsilber J: Protein structure prediction with in-cell photo-crosslinking mass spectrometry and deep learning. Nature Biotechnology 2023; 41(12): 1810-9.
34. Abbas MK, Rassam A, Karamshahi F, Abunora R and Abouseada M: The role of AI in drug discovery. Chembiochem 2024; 25(14): 1-21.
35. Bertoline LM, Lima AN, Krieger JE and Teixeira SK: Before and after AlphaFold2: An overview of protein structure prediction. Frontiers in Bioinformatics 2023; 3: 1120370.
36. Sarkar C, Das B, Rawat VS, Wahlang JB, Nongpiur A, Tiewsoh I, Lyngdoh NM, Das D, Bidarolli M and Sony HT: Artificial intelligence and machine learning technology driven modern drug discovery and development. International Journal of Molecular Sciences 2023; 24(3): 1-40.
37. Du BX, Qin Y, Jiang YF, Xu Y, Yiu SM, Yu H, Shi JY. Compound–protein interaction prediction by deep learning: databases, descriptors and models. Drug Discovery Today 2022; 27(5): 1350-66.
38. Wang DD, Wu W and Wang R: Structure-based, deep-learning models for protein-ligand binding affinity prediction. Journal of Cheminformatics 2024; 16(1): 1-15.
39. Shi W, Yang H, Xie L, Yin XX and Zhang Y: A review of machine learning-based methods for predicting drug–target interactions. Health Information Science and Systems 2024; 12(1): 1-6.
40. Liu Z, Chen Q, Lan W, Lu H and Zhang S: SSLDTI: A novel method for drug-target interaction prediction based on self-supervised learning. Artificial Intelligence in Medicine 2024; 149: 102778.
41. Mouchlis VD, Afantitis A, Serra A, Fratello M, Papadiamantis AG, Aidinis V, Lynch I, Greco D and Melagraki G: Advances in de novo drug design: from conventional to machine learning methods. International Journal of Molecular Sciences 2021; 22(4): 1676.
42. Tang X, Dai H, Knight E, Wu F, Li Y, Li T and Gerstein M: A survey of generative AI for de novo drug design: new frontiers in molecule and protein generation. Briefings in Bioinformatics 2024; 25(4): 338.
43. Bai Q, Liu S, Tian Y, Xu T, Banegas‐Luna AJ, Pérez‐Sánchez H, Huang J, Liu H and Yao X: Application advances of deep learning methods for de novo drug design and molecular dynamics simulation. Wiley Interdisciplinary Reviews: Computational Molecular Science 2022; 12(3): 1581.
44. McGibbon M, Shave S, Dong J, Gao Y, Houston DR, Xie J, Yang Y, Schwaller P and Blay V: From intuition to AI: evolution of small molecule representations in drug discovery. Briefings in bioinformatics 2024; 25(1): 422.
45. Janet JP, Mervin L and Engkvist O: Artificial intelligence in molecular de novo design: Integration with experiment. Current Opinion in Structural Biology 2023; 80: 102575.
46. Askr H, Elgeldawi E, Aboul Ella H, Elshaier YA, Gomaa MM and Hassanien AE: Deep learning in drug discovery: an integrative review and future challenges. Artificial Intelligence Review 2023; 56(7): 5975-6037.
47. Han R, Yoon H, Kim G, Lee H and Lee Y: Revolutionizing medicinal chemistry: the application of artificial intelligence (AI) in early drug discovery. Pharmaceuticals 2023; 16(9): 1259.
48. Liu H, Fan Z, Lin J, Yang Y, Ran T and Chen H: The recent progress of deep-learning-based in-silico prediction of drug combination. Drug Discovery Today 2023; 28(7): 103625.
49. Paul D, Sanap G, Shenoy S, Kalyane D, Kalia K and Tekade RK: Artificial intelligence in drug discovery and development. Drug Discov Today 2021; 26(1): 80-93.
50. Gruber A, Führer F, Menz S, Diedam H, Göller AH and Schneckener S: Prediction of human pharmacokinetics from chemical structure: combining mechanistic modeling with machine learning. Journal of Pharmaceutical Sciences 2024; 113(1): 55-63.
51. Xue HT, Stanley-Baker M, Kong AW, Li HL and Goh WW: Data considerations for predictive modeling applied to the discovery of bioactive natural products. Drug Discovery Today 2022; 27(8): 2235-43.
52. Huang R, Xu M, Zhu H, Chen CZ, Zhu W, Lee EM, He S, Zhang L, Zhao J, Shamim K and Bougie D: Biological activity-based modeling identifies antiviral leads against SARS-CoV-2. Nature Biotechnology 2021; 39(6): 747-53.
53. Jia X, Wen X, Russo DP, Aleksunes LM and Zhu H: Mechanism-driven modeling of chemical hepatotoxicity using structural alerts and an in-vitro screening assay. Journal of Hazardous Materials 2022; 436: 129193.
54. Chokshi N and Chakraborty S: Artificial intelligence from a regulatory perspective: Drug delivery and devices. In A handbook of artificial intelligence in drug delivery. Academic Press 2023; 581-607.
55. Tanoli Z, Vähä-Koskela M and Aittokallio T: Artificial intelligence, machine learning, and drug repurposing in cancer. Expert Opinion on Drug Discovery 2021; 16(9): 977-89.
56. Visan AI and Negut I: Integrating Artificial Intelligence for Drug Discovery in the Context of Revolutionizing Drug Delivery. Life 2024; 14(2): 233.
57. Vora LK, Gholap AD, Jetha K, Thakur RR, Solanki HK and Chavda VP: Artificial intelligence in pharmaceutical technology and drug delivery design. Pharmaceutics 2023; 15(7): 1916.
58. Liu Z, Roberts RA, Lal-Nag M, Chen X, Huang R and Tong W: AI-based language models powering drug discovery and development. Drug Discovery Today 2021; 26(11): 2593-607.
59. Raza MA, Aziz S, Noreen M, Saeed A, Anjum I, Ahmed M and Raza SM: Artificial intelligence (AI) in pharmacy: an overview of innovations. Innovations in pharmacy 2022; 13(2): 1-8.
60. Serrano DR, Luciano FC, Anaya BJ, Ongoren B, Kara A, Molina G, Ramirez BI, Sánchez-Guirales SA, Simon JA, Tomietto G and Rapti C: Artificial intelligence (AI) applications in drug discovery and drug delivery: revolutionizing personalized medicine. Pharmaceutics 2024; 16(10): 1328.
61. Askin S, Burkhalter D, Calado G and El Dakrouni S: Artificial intelligence applied to clinical trials: opportunities and challenges. Health and Technology 2023; 13(2): 203-13.
62. Jayatunga MK, Ayers M, Bruens L, Jayanth D and Meier C: How successful are AI-discovered drugs in clinical trials? A first analysis and emerging lessons. Drug Discovery Today 2024; 104009.
63. Gao M, Liu S, Chen J, Gordon KC, Tian F and McGoverin CM: Potential of Raman spectroscopy in facilitating pharmaceutical formulations development–an AI perspective. International Journal of Pharmaceutics 2021; 597: 120334.
64. Huanbutta K, Burapapadh K, Kraisit P, Sriamornsak P, Ganokratana T, Suwanpitak K and Sangnim T: The artificial intelligence-driven pharmaceutical industry: a paradigm shift in drug discovery, formulation development, manufacturing, quality control, and post-market surveillance. European Journal of Pharmaceutical Sciences 2024; 106938.
65. Bannigan P, Aldeghi M, Bao Z, Häse F, Aspuru-Guzik A and Allen C: Machine learning directed drug formulation development. Advanced Drug Delivery Reviews 2021; 175: 113806.
    

    ---

    How to cite this article:

    Patel P, Hardenia S and Jain DK: Artificial intelligence in drug development. Int J Pharm Sci & Res 2025; 16(5): 1182-93. doi: 10.13040/IJPSR.0975-8232.16(5).1182-93.

    ---

    All © 2025 are reserved by International Journal of Pharmaceutical Sciences and Research. This Journal licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.