DRUG DEVELOPMENT: ROLE OF GENERATIVE ARTIFICIAL INTELLIGENCE

DRUG DEVELOPMENT: ROLE OF GENERATIVE ARTIFICIAL INTELLIGENCE

Skand Arvind, Shivanshi Chauhan and Richa Srivastava *

Amity Institute of Pharmacy, Lucknow, Amity University Uttar Pradesh, Sector 125, Noida, Uttar Pradesh, India.

ABSTRACT: This article explores the transformation of pharmaceutical research and development by the combination of digital twins and generative artificial intelligence (AI), especially in drug discovery. It highlights various cases that illustrate the increasing use of digital twins of biological systems alongside generative AI algorithms to speed up the identification of new drugs. This approach involves simulating vast chemical landscapes and predicting molecular properties, which facilitates the discovery of novel compounds that may have previously been overlooked. By leveraging computational models and machine learning, researchers can design targeted compounds, refine potential therapy candidates, and simulate their interactions within complex biological environments. This method accelerates the identification of promising drug candidates while predicting their efficacy and safety more efficiently, without extensive in-vivo testing. Furthermore, digital twins allow for a more personalized approach to drug development, enhancing the chances of success for specific patient groups. This cutting-edge methodology presents significant opportunities to streamline drug development processes, lower costs, and ultimately improve patient outcomes by delivering more effective treatments. However, the integration of these technologies also poses challenges, including the necessity for interdisciplinary collaboration and ongoing improvements in AI models, computational capabilities, and data integration. As the field progresses, further innovations will be essential to fully harness the potential of these technologies.

Keywords: Drug development, Generative AI, Digital twins, and prospective analysis

INTRODUCTION: Artificial Intelligence (AI) is a vibrant and fast-changing area of computer science focused on creating systems and machines that can perform tasks typically requiring human intelligence. These tasks include learning from experiences, understanding natural language, solving complex problems, making informed decisions, and adapting to new situations. AI aims not only to replicate human cognitive abilities but also to enhance them.

It enables machines to perform tasks more quickly, accurately, and efficiently than humans in many cases. As artificial Intelligence technology progresses, it is transforming industries, society, and our interactions with technology, opening up new avenues for innovation and problem-solving. Artificial Intelligence is often viewed as a broad term that includes several subfields, each offering unique techniques and approaches to building intelligent systems.

Among these, two of the most significant and commonly used subfields are Machine Learning (ML) and Natural Language Processing (NLP). Machine learning (ML) is one branch of artificial intelligence (AI). This group of techniques includes reinforcement learning, unsupervised learning, and supervised learning ². Natural language process [NLP], it is a process in which we know about the interaction between the people and the computers and see how they communicate by using, natural language. It makes it possible for machines to comprehend, decode ³.

Artificial Intelligence comes up with two Sub-fields:

1. Machine learning
2. Deep learning

Machine Learning [ML]: Machine learning program, is the process in which the, algorithmic program is collected for examine data, draw conclusions from it, and decide what to do next. The topic of machine learning is relatively new, incorporating techniques and algorithms that date back several decades, some as far back as the 1960s. The Naïve Bayes classifier and support vector machines are two examples of these traditional methods such as the Naïve Bayes classifier, which works on probabilistic models, and support vector machines (SVMs), known for their use in classification problems, are widely used in data analysis. Apart from classification, cluster analysis algorithms like K-means and tree-based clustering are also available ⁴. Principal component analysis and tSNE, the two techniques used in machine learning, to decrease the dimensionality data and obtain a deeper understanding of its nature.

Deep Learning: A branch of artificial intelligence called "deep learning" uses artificial neural networks as its foundation. Deep learning can be considered a subset of machine learning, as it relies on data to train algorithms that learn patterns and solve complex problems. Deep learning and machine learning are often used interchangeably ^5, but they are distinct concepts. While both fall under the umbrella of artificial intelligence, deep learning specifically utilizes neural networks multi-layered algorithmic structures setting it apart from traditional machine learning approaches.

Areas of Application of AI:

Enhancing Drug Development: AI is being used by the pharmaceutical sector to expedite the discovery and development of new drugs. Conventional drug development is an expensive and time-consuming procedure ⁶. AI accelerates this by predicting the efficacy of specific drug candidates and refining their chemical properties for optimal performance. This not only accelerates development but also reduces costs and improves the likelihood of success in clinical trials ⁷.

AI for Diagnosis: In the area of diagnostics, artificial intelligence has made one of the biggest contributions to healthcare. Deep learning algori-thms and other AI-powered tools are remarkably accurate in analyzing medical images. AI systems, for example, can find anomalies in MRIs, CT scans, and X-rays; these algorithms frequently spot problems that the human eye could miss. When it comes to early diagnosis of diseases like cancer, when early intervention can greatly increase survival chances, this technology is especially helpful ⁸.

Drug Development: The meticulous, multi-phase process of drug development aims to find, test, and introduce novel pharmaceutical products to the market. Enhancing patient outcomes and developing medical therapies depends on this process. An examination of the drug development process identifies numerous critical phases, including regulatory approval, preclinical research, clinical trials, and drug discovery ⁹.

Phases of Discovery: The process starts with the discovery stage, during which researchers find possible therapeutic options. This phase entails learning about the biology of a disease and locating targets, or particular bodily molecules that a medication may interact with. To identify potential molecules, methods such as genetic investigations, computer-aided drug design, and high-throughput screening are frequently used ⁹.

Preclinical Investigations: Upon identification, candidates undergo preclinical research. To assess the compounds' safety, effectiveness, and pharmacokinetics (the way a drug is absorbed, distributed, metabolized, and eliminated), laboratory and animal studies are conducted during this phase ⁷. The goal of preclinical research is to establish if a medication is safe enough to test humans.

Clinical Investigations: There are several stages to conduct clinical trials:

Phase I: Safety, dosage, and side effects are tested on a small number of well-trained volunteers or patients

Phase II: Involves a broader patient population to evaluate the safety and effectiveness of the medication.

Phase III: Enlarges the trial to include a larger patient population in order to verify efficacy, track side effects, and evaluate the medication against standard therapies. To determine the drug's therapeutic value and guarantee its safety for general usage, these trials are crucial for determining the drug's therapeutic value and ensuring its safety for general use ⁸.

Regulatory Acceptance: A drug developer submits a New Drug Application (NDA) or Biologics License Application (BLA) to regulatory bodies, such as the European Medicines Agency (EMA) or the U.S. Food and Drug Administration (FDA), after completing successful clinical trials. Complete clinical trial data are included in the application. Regulatory bodies examine the application, inspect the manufacturing facilities, and approve the drug for marketing if it satisfies all requirements ⁹.

After-Market Monitoring: Phase IV, also known as post-market surveillance, involves monitoring how well the medication works in the general population, spotting any uncommon or long-term side effects, and ensuring continued safety ¹⁰.

AI in Drug Development: The process of developing new drugs is being changed by artificial intelligence (AI), which is making it quicker, more economical, and more efficient. Drug discovery is typically a drawn-out, costly process with a high failure rate. Artificial intelligence (AI) uses cutting-edge algorithms and machine learning approaches to analyze large datasets, find promising drug candidates, and forecast the safety and efficacy of those candidates. Early on in the drug discovery process, artificial intelligence plays a major role in drug development ¹⁹. AI is capable of sorting through enormous volumes of biological data to find interesting molecular structures and forecast how they will interact with particular targets. This expedites the process of identifying promising medication candidates and thereby lowers the time and expense involved in this stage.AI models mimic drug behavior in biological systems during preclinical research. Here, we talk over integration areas, methods and tools for enforcing AI, continuous difficulties, and solutions ¹⁵.

Artificial intelligence (AI) has been used more and more in many spheres of society, most notably the pharmaceutical business. The use of AI in the pharmaceutical industry is highlighted in this review, with applications ranging from drug development and discovery to drug repurposing, increasing pharmaceutical productivity, and clinical trials ^{19, 20}. These applications not only expedite target achievement but also lessen the workload for humans. We also talk about recurring issues, methods, and cross-talk between the AI tools and strategies.

AI in Drug Screening: It takes more than a decade to discover and develop a drug, and this costs about US$2.8 billion on average. Nevertheless, in the field of therapeutic molecules, nine out of ten never pass phase II of clinical trials and regulations ^{31, 32}. According to the synthesis feasibility of the techniques involved, numerous algorithms such as Nearest-Neighbor classifiers, RF, and extreme learning machines, as well as DNNs and SVMs, are employed for VS and are also capable of estimating in vivo activity and toxicity ^{31, 33}. Bayer and Roche, along with IT companies, are among many biopharmaceutical industry players working to build programs for the discovery of treatments for conditions like heart disease and immunopathology ¹⁹.

Estimating the Physical and Chemical Characteristics: The pharmacokinetic properties and the family of target receptors of the drug depend on certain physicochemical characteristics of the drug, such as solubility, partition coefficient, degree of ionization, and intrinsic permeability ³⁴. Various AI-dependent software and models can be used for prospective and predictive analysis of physicochemical properties. For instance, during compound optimization, large datasets are generated for use in training machine learning (ML) programs. These are used in conjunction with molecular descriptors, such as SMILES strings, potential energy measures, electron density of the molecule, and the coordinates of the atoms in 3D, to design feasible molecules through deep neural networks (DNN) that will forecast their properties Zang et al. successfully designed a QSPR framework to identify six physicochemical properties of environmental chemicals from the EPA, known as the EPI Suite ³⁵. The ADMET predictor and ALGOPS program reveal that the developed neural networks are accurate in predicting the lipophilicity/solubility of different compounds ³⁶.

Prediction of Bioactivity: Bioactivity can be predicted using the properties of the compound within the protein binding site. The potency of the drug molecules in relation to the intended protein or receptor is crucial in determining the efficacy of the drug. Drug molecules that do not interact with or bind to the protein being targeted cannot produce the therapeutic effect. Sometimes, developed drug molecules may bind to undesired proteins or receptors, leading to toxicity. Thus, drug-target binding affinity (DTBA) is crucial for predicting the interaction between the drug and target ⁴². On the other hand, similarity in interaction refers to the likeness between a drug and its target, suggesting that they may interact with similar targets. There are web applications that can be used to predict drug-target interactions, where the ‘r’ value ranges from 0 to 1. Among them are Chem Mapper and the Similarity Ensemble Approach (SEA) ⁴³.

Previous ML-based models, such as the Kronecker-regularized least squares (Kron RLS) approach, compare protein molecules and drug molecules to establish the DTBA. "Similarly, SimBoost uses regression trees to estimate DTBA and incorporates features and interaction similarities ⁴⁶. Taking this into account, drug features can be specified using SMILES, LMCS (Large Molecular Complexity Scoring), extended connectivity fingerprints, or a combination of these parameters. DL approaches have been found to perform better than ML because the techniques used do not require structural information of the protein. Some DL methods used to measure DTBA include DeepDTA, PADME, WideDTA, and Deep Affinity ⁴⁶.

The Impact of AI on the Drug Discovery Process: A further critical application domain for AI techniques in drug discovery involves the design of novel compounds with set properties and activity. Traditionally, drug design approaches generally detect existing compounds. Conversely, methods that use [AI] make it possible to create novel compounds with specified properties, and activities quickly and effectively. A deep learning algorithm has recently been taught with a data set of drug compounds featuring known attributes to suggest new therapeutic chemicals,¹⁰ that have prized characteristics demonstrating the capabilities of these techniques in promptly and effectively designing new drug candidates.

DeepMind made, arguable, the most enormous contribution to AI research in the field of biology with the invention of AlphaFold the software platform that will advance the ability of our understanding towards the biology ¹⁹. This is a very strong algorithm that uses protein sequence data together with AI to predict the three-dimensional structures of these proteins. In the field of de novo drug design, ML techniques are being used together with molecular dynamics simulations in such a manner that the efficiency and accuracy of the methodologies are improved. The combination of these methodology is under investigation in order to make the most of synergies that exist between them. DL and IML methods also power this effort. Researchers are using the power of MD, and AI to design the drugs better and also more effective than before.

The concept of rational drug design is defined as the act of modifying existing molecules to achieve specific properties and recognition. Recently, DeepMind developed AlphaFold, a revolutionary software application that enhances biological knowledge. In contrast, the AI-based method allows the design of the new molecules with potential properties and biological activity at a high speed and increased efficiency.

The applicability of these techniques in designing new drug candidates in a short duration is well reflected from the above said exercise where deep learning algorithm was trained with a dataset of compounds and their properties and later used for designing new therapeutic molecules having certain properties like solubility and activity. Recently, DeepMind has produced AlphaFold that is a revolutionary software application to enhance biological knowledge which has brought a significant impact on the AI-related research. It is a powerful tool that can forecast the matching three-dimensional structures of the proteins by the protein sequence data and artificial neural network. This was expected to revolutionize medication discovery and individualized treatment in the domain of structural biology. In the area of structural biology and the biological sciences more broadly, in terms of utilizing AI, AlphaFold is probably a big leap forward.

Machine learning (ML) strategies, together with molecular dynamics (MD) simulations, are currently being used in order to improve speed and reliability in the de novo drug discovery field. With a view to fully leverage these approaches, the option of combining them is being considered under what is referred to as the strategy merge ²⁰. This endeavor also includes an interpretation of deep learning and other specific machine learning algorithms. This way, AI and MD enable researchers to build medications more effectively and efficiently than they could before ²¹.

Successful Drug Discovery Projects that Involved the Use of AI: Numerous case studies prove the effectiveness of the AI model in the framework of drug discovery. For example, Gupta, R., et al ²¹ have recently described the use of AI techniques in the identification of new drugs for the therapy of cancer. These scientists employed a fairly large set that included 166 substances that are reactive to cancer and their corresponding biological activity profile to create an algorithm based on deep learning. Thus, new compounds that hold great potential for cancer treatment in the future were developed, which means that this technology is capable of identifying new treatments.

A new approach proving that ML can be used to identify small-molecule inhibitors of the MEK protein has been presented several weeks ago. MEK itself can be looked upon as a target for cancer therapy but it has been quite challenging to work with efficient inhibitors. In this case, ML was able to uncover the newer inhibitors for this protein ^[22]. Another example is the use of artificial neural networks to screen more inhibitors of beta-site APP cleaving enzyme 1 (BACE1); an enzyme associated with Alzheimer’s disease ²³.

Despite the fact that the cases where AI has been applied to drug design are limited, it is clear that it works efficiently in the search of new antibiotics. A competition of over 100 million molecules was evaluated by an innovative machine learning algorithm that identified antibiotic sorts that are rather powerful – including one that fights many bacteria ^25, including unmanageable ones and tuberculosis.

In the past two years, a novel approaching of using AI to search for drugs to combat against Covid-19 has been explored. Computer-aided reasoning has been applied to analyze large databases of potential compounds to identify those that have the highest likelihood of treating the virus. There are many techniques that have, in some cases, identified potential medications much more quickly than could possibly be achieved through the traditional methods ^{26, 27, 28, 29, 30, 31}. Some other examples which show how such AI based techniques can hasten the drug discovery process as well as help to synthesize better drugs ^{3, 32, 33, 34, 35, 36, 37}.

AI in Pharma Industry: The increasing complexity of manufacturing processes and the growing demand for efficiency and improved product quality are driving modern manufacturing systems to embed human knowledge into machines, transforming manufacturing practices ⁸⁸. The application of AI in manufacturing has the potential to boost the pharmaceutical industry. Tools like CFD use Reynolds-Averaged Navier-Stokes solvers technology to study how agitation and stress levels affect different equipment (e.g., stirred tanks) automating many pharmaceutical operations. Similar systems, such as large eddy simulations and direct numerical simulations, utilize advanced techniques to resolve complicated flow challenges in manufacturing ⁸⁵.

A new computer program supports digital automation for the production and synthesis of molecules, considering a variety of chemical rules and are programmed with the Chemical Assembly scripting language. The synthesis and manufacture of sildenafil, diphenhydramine hydrochloride, and rufinamide have seen yields and purity on par with manual synthesis thanks to its use by scientists ⁸⁹. Technologies powered by AI can project the finish of granulation in granulators that have capacities between 25 and 600. AI technology and neuro-fuzzy logic identified key factors in their responses ^{90 91}. Scientists have used DEM a lot in the drug industry it has been used to study color separation in a double mixer, the effects of blade speed and shape changes, and tablet movement during coating. They've also used it to figure out how long tablets spend under the spray ⁸⁵. AI tools like neural networks and fuzzy models have looked at how machine settings relate to the problem of sticking trying to cut down on tablets sticking during production. Parts of the problem about tablets sticking during production have been addressed by AI tools like neural networks and fuzzy models, which had studied how machine settings are relevant to this issue. These tools are the meta-classifier and tablet- classifier, it uses AI to verify whether your output is finally good ⁹³. They can identify potential pitfalls in the making of that tablet. There is a patent application pending for such an invention that could determine what strength of drug combined with which patch type might work best in any given patient. Part of the solution is a computer that inputs patient information and creates just the right patch for them ⁹⁴.

AI in Quality Control and Quality Assurance: Making the requested product from raw materials involves balancing various factors. Quality control checks on products, plus keeping consistency from batch to batch, need human intervention ⁹³. This approach might not work best in every situation showing why we need to put AI into action at this point. This prompted the FDA to revise Current Good Manufacturing Practices (cGMP) by introducing a 'Quality by Design' approach to define key processes and critical requirements that determine end quality of pharmaceutical product. Decision trees were prepared using primary data, being the combination of manual activity with artificial intelligence algorithm by Gams et al ⁸⁵. They turned these into rules for drivers to examine helping guide future product cycles. Goh and team looked at the dissolution profile, which shows how consistent theophylline tablets are from batch to batch. They used an ANN to predict the dissolution of the tested formula achieving an error rate of 8%. AI may be applied to law in-line production process for the requested quality of product as well. This type of ANN-based monitoring uses a UBG snap-drying process with a tone-adaptive elaboration setting, as well as original search and reverse accumulation. For control purposes in practice, this can be used as a way to predict the temperature and cutlet texture at a later time point (t+Δt) for a fixed set of operating conditions ⁹⁶.

An Electronic Lab Tablet which reads an automated data entry platform can then under tight control guards maximize protection for the product a repeating analysis--smart ways like these should mean good-housekeeping holds up pretty well! Furthermore, discovery techniques-based data mining and visualization to Total Quality Management 990 ⁹⁷. And intelligent opinions, and eventually building new technology into those two things - these system intelligence-controlled decisions quality control precious ⁹⁹.

AI in Clinical Trial Design: Clinical trials Clinical trials of a new investigational medicinal product take an average of six to seven years to scientifically establish its safety and efficacy in treating a specific illness in people, not to mention the large sum of money required. However, only one in ten of the molecules that go through these trials manage to get clearance, an outcome that is staggering for the business. Such failures could be attributed to poor patient selection, absence of technical standards, and poor facilities ¹⁰⁰. However, by employing the AI system, such problems are considerably reduced because there is tremendous digital medical data available at the moment. Recruitment of patients occupies a third of the trial’s time ¹⁰¹.

The selection of the right patients is a key determinant of the success of a clinical study, because it avoids roughly 86% of failure scenarios ¹⁰². By means of patient-specific genome–exposome profile analysis, the AI is capable of recruiting only the particular diseased population in the [phase II and phase III clinical trial]. It can contribute to the premature division of the landscapes of therapeutic targets in the chosen patients by design ^{19, 101}. Other AI based approaches such as predictive machine learning and other forms of reasoning help in preclinical molecular discovery and also lead optimization such that the resultant drugs are more likely to pass clinical trials with respect to the chosen patient population ¹⁰¹. Patient dropout is identified as a major reason for failure in 30% of clinical trials because the number of patients to be recruited is increased to meet the required number in order to complete the trial thus incurring a lot of time and money. This can be prevented when patient is closely observed and can be helped to observe the planned protocol for the research trial ¹⁰². Patients that are diagnosed with schizophrenia are commonly prescribed antipsychotic medications and so Ai Cure developed mobile applications that ensure that the identified patients take the required doses as required by the Phase II trial. This led to a 25% compliance percentage among the patients so that the clinical trial was successfully accomplished ¹⁹.

Artificial Intelligence in the Management of Pharmaceutical Products:

Marketing Positioning through AI: Any business strategy for an organization that is looking to develop a unique brand must include market positioning, which is the process of endowing a product with a personality in the market to attract customers ^{103, 104}. This marketing approach was taken up by the producers of original Viagra aiming at treating other life issues alongside erectile problems among men. The platform that technology and e-commerce provide has made it easy for companies to have its organic recognition as a brand name in public domain ¹⁰⁶. Businesses use search engines as one of the technology platforms to take center stage in online marketing. Businesses are always looking for ranking their websites higher than those of their competitors so that they can establish their brand first. There were also other tools like; particle swarm optimization algorithms, and statistical analysis techniquesis introduced by Kennedy and Eberhart in 1995 coupled with (NNs) that helped to understand markets better. This way, it is possible to accurately predict consumer demand and provide a basis for a marketing plan of a product ¹⁰⁷.

AI-Based Advanced Applications:

AI-based Nanorobots for Drug Delivery: AI-based cutting-edge applications, such as nanorobots for drug delivery, involve nanorobots with integrated circuits, detectors, power sources, and data backup systems. AI technologies maintain these components ^{117, 118}.

Engineers program them to avoid collisions, identify targets, detect and attach, and leave the body. New nano/microrobots can navigate to specific areas based on body conditions like ph. This improves their effectiveness and reduces side effects. To develop implantable nanorobots for drug and gene delivery, scientists must consider drug adaptation long-term release ¹¹⁹. AI tools like; integrators and, neural networks fuzzy logic control the drug release. Doctors use microchip implants to program drug release and locate the implant within the body.

Artificial Intelligence used for Forecasting Synergism/Antagonism alongside Combination Drug Delivery: AI in combination drug delivery and symbiosis/antagonism prediction Many drug combinations get approval and hit the market to treat complex conditions, like TB and cancer, because they can boost each other for faster healing ^{120, 121}. Picking the right drugs that work well together means testing lots of drugs, which takes a long time; for example, cancer treatment often uses six or seven drugs together. ANNs logistic regression, and network models help screen drug combos and make treatments more effective ^120,122.

Rashid and his team created a system to find the best drug combo to treat multiple myeloma that doesn't respond to bortezomib. They looked at 114 FDA approved drugs. Their model said decitabine (Dec) and mitomycin C (MitoC) work best as a pair, while Dec, MitoC, and mechlorethamine make the top three-drug mix ¹²¹. Combo drug delivery can work even better with info on how drugs help. Li and his team designed a synergistic drug combination model, made possible through RF, to predict synergistic combinations of anticancer drugs. The approach was successful in predicting 28 synergistic anticancer combinations based on gene expression profiles and various networks. In their report, Mason and colleagues discussed three examples of these combinations, though the remaining ones may still carry importance ¹²³.

The advent of AI in Nano-medicine: The advent of AI in nanomedicine Nanomedicines are those forms of medicines that employ drug and nanotechnology together to alleviate, treat or diagnose and monitor complex diseases like HIV, cancer, malaria, asthma or other inflammatory diseases. Due to the improved efficiency and effective treatment as a result of nanoparticle modified drug delivery systems, this mode of delivery has recently gained prominence in therapeutics and diagnostics ^{121, 124}. There are several challenges in formulation development ¹²⁵. Perhaps here astonishment might lie because of AI and nanotech together. Clothes employ any stage of contacts of the field to spot regions with certain properties that could induce the formation of cluster shapes ⁸³. Coarse-grained simulation provides aid to the estimation of drug encapsulation in the dendrimers and the estimation of drug – dendrimer interaction alongside chemical computation. The influence of surface chemistry on the internalization of nanoparticles into cells could also have been investigated by using computer programs such as LAMMPS and GROMACS 4 ⁸³. This increased silicasome intake three- to four-fold because iRGD enhances silica some transcytosis, hence leading to better treatment outcomes and increased overall survival ¹²⁴.

AI's Pharmaceutical Market: AI's pharmaceutical market, pharma companies increasingly look at AI as a means to reduce the costly burden of VS along with its high failure rates. The AI market that was valued at US$ 200 million in, (2015) reached (US$700 million in 2018). It is expected to increase by 40% from 2017 to 2024, meaning that it is bound to make radical changes in pharmaceuticals and medical sectors. Most pharmaceutical firms have invested in AI and are still investing. They have also partnered with AI startups to provide essential healthcare technologies.

Constant Challenges to the use of AI: The effectiveness of artificial intelligence depends on the volume and quality of available data, as data is essential for training these systems effectivity. If a company is in need of access to data from a number of database providers, then it will have to pay more, and for correct result prediction, the data also needs to be reliable and of high quality. Other challenges that further impede the full-scale implementation of artificial intelligence in the pharma sector include a lack of trained personnel to run the AI-based platforms, small organizations having small budgets, fears that such replacement will lead to a loss of jobs, skepticism regarding data emanating ⁶. With time, some of the jobs in clinical trials, manufacturing process, and, supply chains, and research of medication and sales will become automated; however, all those jobs fall into the category of "narrow AI," which simply means for the AI to be appropriate for a particular task, it needs to be trained on a huge amount of data first. Hence, for this AI platform to be developed, implemented, and run successfully, the requirement for human intervention becomes necessary. But since the reins of all the mundane tasks have already been taken over by AI itself, leaving space for human intelligence to be applied to higher insights and creative expression, the fear of unemployment might not have been that real after all.

Yet, certain pharmaceutical companies have embraced advancements in AI technology, with estimates of US$2.199 billion in revenues to be earned from AI solutions by the pharmaceutical industry by the year 2022 with investments of US$7.20 billion over more than 300 deal strategies that were undertaken from the year 2013 to the year 2018. Post-deployment, pharmaceutical companies often remain uncertain about the extent to which AI can address the concerns it was implemented for and the limits of its capabilities ¹²⁷. To properly capitalize on the AI platform, appropriate resources inside an organization, such as successful data scientists, suitable software engineers, if possible, with fundamentals of AI knowledge and clear understanding business objectives and R weekends of the firm.

CONCLUSION AND FUTURE DIRECTIONS: The advanced capabilities of AI aim to reduce the challenges faced by pharmaceutical firms, which impact the entire product life cycle and medication development process. This could be one of the reasons for the rise in start-ups within this domain. The increasing cost of drugs and therapies is one of the major challenges currently faced by the healthcare sector. Therefore, society needs to adopt an entirely new approach to address this challenge ²³. AI-enabled production of pharmaceutical products allows for drug personalization based on each patient's specific needs by adjusting the dosage, release parameters, and other critical factors ^128. State-of-the-art AI-based technologies will further accelerate product market entry, improve product quality, enhance production safety, and reduce costs. These are the advantages that have accorded importance to automation. A major concern regarding the adoption of AI innovations is the potential job losses and the strict regulations surrounding AI integration. However, these innovations are not designed to displace humans but to simplify and optimize work processes ¹²⁹. AI can accelerate and streamline the identification of hit compounds, propose synthesis pathways, predict the required chemical structures, and provide better insights into drug-target interactions and structure-activity relationships (SAR). AI can significantly enhance drug development, optimization, and integration into appropriate dosage forms additionally; AI enables timely decision-making, accelerates high-quality production, and ensures batch-to-batch consistency. Comprehensive market analysis and prediction enabled by AI can validate product safety and efficacy during clinical trials while optimizing market placement and pricing strategies. AI is not yet fully commercialized, and while it must overcome several obstacles before widespread adoption, it is highly likely that the pharmaceutical sector will find it indispensable in the near future.

ACKNOWLEDGEMENTS: Nil

CONFLICTS OF INTEREST: Nil

REFERENCES:

1. Turing AM: Computing Machinery and Intelligence. In The Essential Turing; Oxford Academic: Oxford, UK, 1950; 59: 433–460.
2. Ramesh A: Artificial intelligence in medicine. Ann R Coll. Surg Engl 2004; 86: 334–338.
3. Miles J and Walker A: The potential application of artificial intelligence in transport. IEE Proc.-Intell. Transport Syst 2006; 153: 183–198.
4. Xu Y, Liu X, Cao X, Huang C, Liu E, Qian S, Liu X, Wu Y, Dong F and Qiu CW: Artificial intelligence: A powerful paradigm for scientific research. Innovation 2021; 2: 100179.
5. Zhuang D and Ibrahim AK: Deep learning for drug discovery: A study of identifying high efficacy drug compounds using a cascade transfer learning approach. Appl Sci 2021; 11: 7772.
6. Pu L, Naderi M, Liu T, Wu HC, Mukhopadhyay S and Brylinski M: EToxPred: A machine learning-based approach to estimate the toxicity of drug candidates. BMC Pharmacol Toxicol 2019; 20: 2.
7. Lamberti MJ: A study on the application and use of artificial intelligence to support drug development. Clin Ther 2019; 41: 1414–1426.
8. Karara AH, Edeki T and McLeod J: PhRMA survey on the conduct FDA, The FDA and the Drug Development Process: How the FDA ensures that drugs are safe and effective, FDA Fact sheet 2002.
9. Adams CP and Brantner VV: New Drug Development: Estimatingentry from human clinical trials. Bureau of Economics Federal Trade commission 2003.
10. Steels L and Brooks R: Routledge; The Artificial Life Route to Artificial Intelligence: Building Embodied, Situated Agents 2018.
11. Gómez-Bombarelli R, Wei JN, Duvenaud D, Hernández-Lobato JM, Sánchez-Lengeling B, Sheberla D, Aguilera-Iparraguirre J, Hirzel TD, Adams RP and Aspuru-Guzik A: Automatic Chemical Design Using a Data-Driven Continuous Representation of Molecules. ACS Central Sci 2018; 4: 268–276.
12. Kalyane D: Artificial intelligence in the pharmaceutical sector: current scene and future prospect. In: Tekade Rakesh K., editor. The Future of Pharmaceutical Product Development and Research. Elsevier 2020; 73–107.
13. Da Silva IN: Springer; 2017. Artificial Neural Networks.
14. Medsker L and Jain LC: CRC Press; Recurrent Neural Networks: Design and Applications 1999.
15. Hänggi M and Moschytz GS: Springer Science & Business Media. Cellular Neural Networks: Analysis, Design and Optimization 2000.
16. Rouse M: 2017 IBM Watson Supercomputer. Accessed 13 October 2020.
17. Vyas M: Artificial intelligence: the beginning of a new era in pharmacy profession. Asian J Pharm 2018; 12: 72–76.
18. Duch W: Artificial intelligence approaches for rational drug design and discovery. CPD 2007; 13: 1497–1508.
19. Blasiak A: CURATE. AI: optimizing personalized medicine with artificial intelligence. SLAS Technol 2020; 25: 95–105.
20. Nussinov R, Zhang M, Liu Y and Jang H: AlphaFold, Artificial Intelligence (AI), and Allostery. J Phys Chem B 2022; 126: 6372–6383. doi: 10.1021/acs.jpcb.2c04346
21. 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 Interdiscip Rev Comput Mol Sci 2022; 12: 1581.
22. Gupta R, Srivastava D, Sahu M, Tiwari S, Ambasta RK and Kumar P: Artificial intelligence to deep learning: Machine intelligence approach for drug discovery. Mol Divers 2021; 25: 1315–1360.
23. Zhu J, Wang J, Wang X, Gao M, Guo B, Gao M, Liu J, Yu Y, Wang L and Kong W: Prediction of drug efficacy from transcriptional profiles with deep learning. Nat Biotechnol 2021; 39: 1444–1452.
24. Dhamodharan G and Mohan CG: Machine learning models for predicting the activity of AChE and BACE1 dual inhibitors for the treatment of Alzheimer ’s disease Mol Divers 2022; 26: 1501–1517.
25. Melo MCR, Maasch JRMA and de la Fuente-Nunez C: Accelerating antibiotic discovery through artificial intelligence. Commun Biol 2021; 4: 1050.
26. Marchant J: Powerful antibiotics discovered using AI. Nature Online ahead of print 2020.
27. Lv H, Shi L, Berkenpas JW, Dao FY, Zulfiqar H, Ding H, Zhang Y, Yang L and Cao R: Application of artificial intelligence and machine learning for COVID-19 drug discovery and vaccine design. Brief. Bioinform 2021; 22: 320.
28. Monteleone S, Kellici TF, Southey M, Bodkin MJ and Heifetz A: Methods in Molecular Biology. Volume 2390. Humana Press Inc.; Totowa, NJ, USA: Fighting COVID-19 with Artificial Intelligence 2022; 103–112.
29. Zhou Y, Wang F, Tang J, Nussinov R and Cheng F: Artificial intelligence in COVID-19 drug repurposing. Lancet Digit Health 2020; 2: 667–e676.
30. Verma N, Qu X, Trozzi F, Elsaied M, Karki N, Tao Y, Zoltowski B, Larson EC and Kraka E: Predicting potential Sars-Cov-2 drugs-in depth drug database screening using deep neural network framework ssnet, classical virtual screening and docking. Int J Mol Sci 2021; 22: 1392.
31. Bung N, Krishnan SR, Bulusu G and Roy A: De novo design of new chemical entities for SARS-CoV-2 using artificial intelligence. Future Med Chem 2021; 13: 575–585.
32. Floresta G, Zagni C, Gentile D, Patamia V and Rescifina A: Artificial Intelligence Technologies for COVID-19 De Novo Drug Design. Int J Mol Sci 2022; 23: 3261.
33. Vatansever S, Schlessinger A, Wacker D, Kaniskan HÜ, Jin J, Zhou MM and Zhang B: Artificial intelligence and machine learning-aided drug discovery in central nervous system diseases: State-of-the-arts and future directions. Med Res Rev 2021; 41:1427–1473. doi: 10.1002/med.21764.
34. Farghali H, Canová NK and Arora M: The Potential Applications of Artificial Intelligence in Drug Discovery and Development. Physiol Res 2021; 70(4): 715–722. doi: 10.33549/physiolres.934765.
35. Ganesh GS, Kolusu AS, Prasad K, Samudrala PK and Nemmani KVS: Advancing health care via artificial intelligence: From concept to clinic. Eur J Pharmacol 2022; 934.
36. Koromina M, Pandi MT and Patrinos GP: Rethinking drug repositioning and development with artificial intelligence, machine learning, and omics. OMICS A J Integr Biol 2019; 23: 539–548.
37. Mak KK and Pichika MR: Artificial intelligence in drug development: Present status and future prospects. Drug Discov. Today 2019; 24: 773–780. doi: 10.1016/j.drudis.2018.11.014.
38. Fleming N: How artificial intelligence is changing drug discovery spotlight. Nature 2018; 557: 55-57.
39. Lusci A: Deep architectures and deep learning in chemoinformatics: the prediction of aqueous solubility for drug-like molecules. JCIM 2013; 53: 1563–1575.
40. Kumar R: Prediction of human intestinal absorption of compounds using artificial intelligence techniques. Curr Drug Discovery Technol 2017; 14: 244–254.
41. Rupp M: Estimation of acid dissociation constants using graph kernels. Mol Inf 2010; 29: 731–740.
42. Chai S: A grand product design model for crystallization solvent design. Comput Chem Eng 2020; 135: 106764.
43. Thafar M: Comparison study of computational prediction tools for drug–target binding affinities. Frontiers Chem 2019; 7: 1–19.
44. Öztürk H: DeepDTA: deep drug–target binding affinity prediction. Bioinformatics 2018; 34: 821–829.
45. Lounkine E: Large-scale prediction and testing of drug activity on side-effect targets. Nature. 2012; 486:361–367.
46. Mahmud S.H. iDTi-CSsmoteB: identification of drug–target interaction based on drug chemical structure and protein sequence using XGBoost with over-sampling technique SMOTE. IEEE Access 2019; 7: 48699–48714.
47. Gao KY: Interpretable drug target prediction using deep neural representation. In: Lang Jérôme., editor. Proceedings of the Twenty-Seventh. International Joint Conference on Artificial Intelligence 2018; 3371–3377.
48. Feng Q. Padme: a deep learning-based framework for drug–target interaction prediction. arXiv 2018 arXiv:1807.09741.
49. Karimi M: DeepAffinity: interpretable deep learning of compound–protein affinity through unified recurrent and convolutional neural networks. Bioinformatics 2019; 35(18): 3329–3338.
50. Pu L: eToxPred: a machine learning-based approach to estimate the toxicity of drug candidates. BMC Pharmacol Toxicol 2019; 20: 2.
51. Mayr A: DeepTox: toxicity prediction using deep learning. Frontiers Environ Sci 2016; 3: 80.
52. Basile AO, Yahi A and Tatonetti NP: Artificial intelligence for drug toxicity and safety. Trends Pharmacol Sci 2019; (9): 624–635. doi: 10.1016/j.tips.2019.07.005. Epub 2019 Aug 2. PMID: 31383376; PMCID: PMC6710127.
53. Lysenko A: An integrative machine learning approach for prediction of toxicity-related drug safety. Life Sci Alliance 2018; 1.
54. Basile AO: Artificial intelligence for drug toxicity and safety. Trends Pharmacol Sci 2019; 40: 624–635.
55. Gayvert KM: A data-driven approach to predicting successes and failures of clinical trials. Cell Chem Biolo 2016; 23: 1294–1301.
56. Jimenez-Carretero D: Tox_(R) CNN: deep learning-based nuclei profiling tool for drug toxicity screening. PLoS Comput Biol 2018; 14.
57. Wan F and Zeng J: Deep learning with feature embedding for compound–protein interaction prediction. bioRxiv. 2016; 2016.
58. AlQuraishi M: End-to-end differentiable learning of protein structure. Cell Syst 2019; 8: 292–301.
59. Hutson M: AI protein-folding algorithms solve structures faster than ever. Nature 2019; XX: YYY–ZZZ.
60. Avdagic Z: Artificial intelligence in prediction of secondary protein structure using CB513 database. Summit Transl. Bioinf 2009; 2009: 1.
61. Tian K. Boosting compound-protein interaction prediction by deep learning. Methods. 2016; 110:64–72.
62. Wang F: Computational screening for active compounds targeting protein sequences: methodology and experimental validation. J Chem Inf Model 2011; 51: 2821–2828.
63. Yu H: A systematic prediction of multiple drug–target interactions from chemical, genomic, and pharmacological data. PLoS One 2012; 7: 37608.
64. Xiao X. iDrug-Target: predicting the interactions between drug compounds and target proteins in cellular networking via benchmark dataset optimization approach. J Biomol Struct Dyn 2015; 33: 2221–2233.
65. Persidis A: The benefits of drug repositioning. Drug Discov World 2011; 12: 9–12.
66. Koromina M: Rethinking drug repositioning and development with artificial intelligence, machine learning, and omics. Omics 2019; 23: 539–548.
67. Park K: A review of computational drug repurposing. Transl Clin Pharmacol 2019; 27: 59–63.
68. Zeng X: Target identification among known drugs by deep learning from heterogeneous networks. Chem Sci 2020; 11: 1775–1797.
69. Achenbach J: Computational tools for polypharmacology and repurposing. Fut Med Chem 2011; 3: 961–968.
70. Yi-Yu K: Artificial intelligence approach fighting COVID-19 with repurposing drugs. Biomed J 2020; 43: 355–362.
71. Li X: Prediction of synergistic anticancer drug combinations based on drug target network and drug induced gene expression profiles. Artif Intell Med 2017; 83: 35–43.
72. Reddy AS and Zhang S: Polypharmacology: drug discovery for the future. Expert Rev Clin Pharmacol 2013; 6: 41–47.
73. Li Z and Kinome X: a web application for predicting kinome-wide polypharmacology effect of small molecules. Bioinformatics 2019; 35: 5354–5356.
74. Cyclica; Cyclica Launches Ligand Express™, a Disruptive Cloud–Based Platform to Revolutionize Drug Discovery 2017.
75. Corey E and Wipke WT: Computer-assisted design of complex organic syntheses. Science. 1969; 166:178–192.
76. Grzybowski B.A. Chematica: a story of computer code that started to think like a chemist. Chem 2018; 4: 390–398.
77. Klucznik T: Efficient syntheses of diverse, medicinally relevant targets planned by computer and executed in the laboratory. Chem 2018; 4: 522–532.
78. Segler MH: Planning chemical syntheses with deep neural networks and symbolic AI. Nature 2018; 555: 604–610.
79. Putin E. Reinforced adversarial neural computer for de novo molecular design. J Chem Inform Modeling 2018; 58: 1194–1204.
80. Segler MH: Generating focused molecule libraries for drug discovery with recurrent neural networks. ACS Cent Sci 2018; 4: 120–131.
81. Popova M: Deep reinforcement learning for de novo drug design. Sci Adv 2018; 4: 7885.
82. Merk D: De novo design of bioactive small molecules by artificial intelligence. Mol Inf 2018; 37: 1700153.
83. Schneider G and Clark DE: Automated de novo drug design: are we nearly there yet? A new Chem 2019; 131: 10906–10917.
84. Guo M: A prototype intelligent hybrid system for hard gelatin capsule formulation development. Pharm Technol 2002; 6: 44–52.
85. Mehta CH: Computational modeling for formulation design. Drug Discovery Today 2019; 24: 781–788.
86. Zhao C: Toward intelligent decision support for pharmaceutical product development. J Pharm Innovation 2006; 1: 23–35.
87. Rantanen J and Khinast J: The future of pharmaceutical manufacturing sciences. J Pharm Sci 2015; 104: 3612–3638.
88. Ketterhagen WR: Process modeling in the pharmaceutical industry using the discrete element method. J Pharm Sci 2009; 98: 442–470.
89. Chen W: Mathematical model-based accelerated development of extended-release metformin hydrochloride tablet formulation. AAPS Pharm Sci Tech 2016; 17: 1007–1013.
90. Meziane F: Intelligent systems in manufacturing: current developments and future prospects. Integr Manuf Syst 2000; 11: 218–238.
91. Steiner S: Organic synthesis in a modular robotic system driven by a chemical programming language. Science 2019; 363: 2211.
92. Faure A: Process control and scale-up of pharmaceutical wet granulation processes: a review. Eur J Pharm Biopharm 2001; 52: 269–277.
93. Landin M: Artificial intelligence tools for scaling up of high shear wet granulation process. J Pharm Sci 2017; 106: 273–277.
94. Das MK and Chakraborty T: ANN in pharmaceutical product and process development. In: Puri Munish., editor. Artificial Neural Network for Drug Design, Delivery and Disposition. Elsevier 2016; 277–293.
95. Gams M: Integrating artificial and human intelligence into tablet production process. AAPS Pharm Sci Tech 2014; 15: 1447–1453.
96. Kraft DL: System and methods for the production of personalized drug products. US20120041778A1.
97. Aksu B: A quality by design approach using artificial intelligence techniques to control the critical quality attributes of ramipril tablets manufactured by wet granulation. Pharm Dev Technol 2013; 18: 236–245.
98. Goh WY: Application of a recurrent neural network to prediction of drug dissolution profiles. Neural Comput Appl 2002; 10: 311–317.
99. Drăgoi EN: On the use of artificial neural networks to monitor a pharmaceutical freeze-drying process. Drying Technol 2013; 31: 72–81.
100. Reklaitis R: PharmaHub; Towards Intelligent Decision Support for Pharmaceutical Product Development 2008.
101. Wang X: International Conference on Computational Intelligence and Software Engineering. IEEE. Intelligent quality management using knowledge discovery in databases 2009; 1–4.
102. Hay M: Clinical development success rates for investigational drugs. Nat Biotechnol 2014; 32: 40–51.
103. Harrer S: Artificial intelligence for clinical trial design. Trends Pharmacol Sci 2019; 40: 577–591.
104. Fogel DB: Factors associated with clinical trials that fail and opportunities for improving the likelihood of success: a review. Contemp. Clin. Trials Commun 2018; 11: 156–164.
105. Kalafatis SP: Positioning strategies in business markets. J. Bus. Ind. Marketing 2000; 15: 416–437.
106. Jalkala AM and Keränen J: Brand positioning strategies for industrial firms providing customer solutions. J Bus Ind Marketing 2014; 29: 253–264.
107. Ding M: Springer. Innovation and Marketing in the Pharmaceutical Industry 2016.
108. Dou W: Brand positioning strategy using search engine marketing. Mis Quarterly 2010; 261–279.
109. Chiu CY: An intelligent market segmentation system using k-means and particle swarm optimization. Expert Syst Appl 2009; 36: 4558–4565.
110. Toker D: A decision model for pharmaceutical marketing and a case study in Turkey. Ekonomska Istraživanja 2013; 26: 101–114.
111. Singh J: Sales profession and professionals in the age of digitization and artificial intelligence technologies: concepts, priorities, and questions. J Pers Selling Sales Manage 2019; 39: 2–22.
112. Milgrom PR and Tadelis S: National Bureau of Economic Research; How Artificial Intelligence and Machine Learning Can Impact Market Design 2018.
113. Davenport T: How artificial intelligence will change the future of marketing. J Acad Marketing Sci 2020; 48: 24–42.
114. Syam N and Sharma A: Waiting for a sales renaissance in the fourth industrial revolution: machine learning and artificial intelligence in sales research and practice. Ind Marketing Manage 2018; 69: 135–146.
115. Mahajan KN and Kumar A: Business intelligent smart sales prediction analysis for pharmaceutical distribution and proposed generic model. Int J Comput Sci Inform Technol 2017; 8: 407–412.
116. Duran O: Neural networks for cost estimation of shell and tube heat exchangers. Expert Syst Appl 2009; 36: 7435–7440.
117. Park Y: A literature review of factors affecting price and competition in the global pharmaceutical market. Value Health 2016; 19: 265.
118. de Jesus A. Emerj; AI for Pricing – Comparing 5 Current Applications 2019.
119. Hassanzadeh P: The significance of artificial intelligence in drug delivery system design. Adv Drug Delivery Rev 2019; 151: 169–190.
120. Luo M: Micro‐/nanorobots at work in active drug delivery. Adv Funct Mater 2018; 28: 1706100.
121. Fu J and Yan H: Controlled drug release by a nanorobot. Nat. Biotechnol 2012; 30: 407–408.
122. Calzolari D: Search algorithms as a framework for the optimization of drug combinations. PLoS Comput Biol 2008; 4: 1000249.
123. Wilson B and KM G: Artificial intelligence and related technologies enabled nanomedicine for advanced cancer treatment. Future Med 2020; 15: 433–435.
124. Tsigelny IF: Artificial intelligence in drug combination therapy. Brief Bioinform 2019; 20: 1434–1448.
125. Mason DJ: Using machine learning to predict synergistic antimalarial compound combinations with novel structures. Front Pharmacol 2018; 9: 1096.
126. Ho D: Artificial intelligence in nanomedicine. Nanoscale Horiz 2019; 4: 365 377.
127. Sacha GM and Varona P: Artificial intelligence in nanotechnology. Nanotechnology 2013; 24: 452002.
128. Pellat G and Anghelache C: (Year) Governance in the EU Member States in the Era of Big Data, Publisher.
129. Research and Markets. Research and Markets. Global Growth Insight - Role of AI in the Pharmaceutical Industry Exploring Key Investment Trends, Companies-to-Action, and Growth Opportunities for AI in the Pharmaceutical Industry 2019; 2018-2022.
130. Jämsä-Jounela SL: Future trends in process automation. Annu Rev Control 2007; 31: 211–220.
131. Davenport TH and Ronanki R: Artificial intelligence for the real world. Harvard Bus Rev 2018; 96: 108–116.
     

     ---

     How to cite this article:

     Arvind S, Chauhan S and Srivastava R: Drug development: role of generative artificial intelligence. Int J Pharm Sci & Res 2025; 16(7): 1914-25. doi: 10.13040/IJPSR.0975-8232.16(7).1914-25.

     ---

     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.