AI and The Future of Drug Discovery: From Innovation to Implementation

Rutuja R. Kamble,* Sanika S. Varale, Hrushikesh D. Sajanikar,   Shivani S. Kavale, Sheetal K. Kamble, Shoan V. Mane

Y. D. Mane Institute of Pharmacy, Kagal 416 216, Maharashtra, India

*Correspondence: rutujak725@gmail.com;

DOI: https://doi.org/10.71431/IJRPAS.2025.41002  

INTRODUCTION

Artificial Intelligence (AI) is derived from two words: Artificial, meaning "man-made," and Intelligence, referring to "the ability to think." It is a field of computer science that enables the creation of machines capable of imitating human behaviour and making decisions. AI is said to exist when machines demonstrate human-like abilities such as learning, reasoning, and problem-solving. Essentially, AI equips computer programs with the capacity to think and learn independently. In simple terms, it represents the simulation of human intelligence within machines to perform tasks that typically require human effort [1, 2]

In recent years, AI has gained significant traction across multiple sectors of society, including the pharmaceutical industry [3]. Within drug discovery and development, AI is proving to be a valuable asset. Traditionally, drug discovery has been a lengthy, costly, and highly complex process, often taking more than a decade to progress from identifying a molecule to obtaining regulatory approval and launching the drug in the market. Each stage carries a considerable risk of failure, and the majority of drug candidates never reach commercialization. This makes the conventional process both inefficient and financially demanding [4].

AI, however, offers the potential to revolutionize this pipeline by making drug discovery faster, more cost-effective, and capable of producing more successful outcomes. Its integration into pharmaceutical research is transforming the industry by accelerating the development of new drugs while enhancing their efficacy and safety. Nonetheless, the effective use of AI depends on access to high-quality data, careful consideration of ethical issues, and an awareness of the inherent limitations of AI-driven methods [5].

In drug discovery, Machine Learning (ML), Deep Learning (DL), and Large Language Models (LLMs) have emerged as powerful tools, enabling the analysis of complex biological datasets, prediction of drug–target interactions, de novo molecular design, and optimization of synthetic pathways. This paradigm shift from hypothesis-driven to data-driven approaches highlights the critical role of AI as an enabler of next-generation therapeutics [6].

Figure 1 : Schematic representation of the main stages during the drug discovery and drug development process. The star represents those stages where AI plays a key role in pharmaceutical processes.

APPLICATION OF AI IN DRUG DISCOVERY:

AI is revolutionizing drug discovery by accelerating the process, reducing costs, and improving accuracy through various applications like identifying novel drug targets, predicting drug properties, designing new molecules, virtual screening, and drug repurposing. AI algorithms analyze vast datasets to find patterns that traditional methods miss, leading to faster development of more effective and safer medicines [7].

1. Drug Design

AI is transforming drug design by rapidly identifying promising lead compounds through advanced analysis of molecular structures and predicting their binding potential. This speeds up the drug development process. While traditional computational methods face challenges like high costs and long processing times, AI overcomes these issues, improving efficiency and reliability in discovering effective, safe, and patentable drugs.

A major focus of drug design is targeting protein structures, as many diseases stem from protein dysfunction. Traditionally, predicting 3D protein structures was slow, costly, and often inaccurate. AI, especially deep learning, has revolutionized this area by accurately predicting protein structures and interactions. This advancement boosts structural drug design, making drug development faster, more cost-effective, and more successful [8].

2. Target Identification

Identifying suitable drug targets is a crucial step in drug discovery, as it helps define the biological pathways and mechanisms to achieve therapeutic outcomes. Machine learning plays a key role in this process by analyzing large and complex datasets—such as genomic, proteomic, and clinical data—to identify and prioritize potential disease-associated targets. Techniques like PCA and t-SNE help reveal hidden relationships between biological entities, aiding in the identification of promising drug targets.

Machine learning algorithms enhance drug target identification by integrating diverse data sources to assess drug ability, safety, and therapeutic relevance. Tools like the Drug–Gene Interaction Database (DGIdb) use machine learning to curate drug–gene interactions, while the Connectivity Map (CMap) analyses gene expression profiles from drug-treated cells to identify targets based on transcriptional signatures. These signatures reveal functional connections between proteins, drugs, and pathways, offering a valuable reference for drug discovery [9, 10].

3. Lead Optimization

Lead optimization aims to enhance the potency, selectivity, and pharmacokinetic properties of drug candidates through chemical modifications. Traditionally time-consuming and inefficient, this process has been improved by machine learning, which enables accurate prediction of biological activity and drug-like properties.

Machine learning models, such as QSAR and GANs, learn from large datasets to uncover structure–activity relationships (SARs), guiding rational compound design. Tools like DeepChem and Schrödinger’s Maestro platform use deep learning and molecular docking to predict compound efficacy and prioritize leads, streamlining drug development with greater speed and precision. [11, 12, 7]

4. Structure–Activity Relationship (SAR) Modelling

AI models can create connections between a compound's chemical makeup and biological activity. This empowers scientists to create molecules with desired properties, like high potency, selectivity, and advantageous pharmacokinetic profiles, in order to optimize drug candidates [13]

5. De Novo Drug Design

De novo drug design involves creating novel molecules with desired pharmacological properties, but faces challenges due to the vastness of chemical space. Approaches are broadly ligand-based (rule-based or rule-free), structure-based, or hybrids:

Rule-based methods use predefined building blocks and reaction rules, ensuring synthesizability but limiting diversity.

Rule-free methods, often powered by deep learning (RNNs, VAEs, GANs, graph models), generate molecules directly from data, enabling broader exploration but risking impractical designs.

Hybrid approaches combine both, balancing novelty with synthetic feasibility.

Recent advances in generative neural networks have accelerated ligand-based de novo design, with platforms like MOSES and GuacaMol enabling standardized benchmarking. Current challenges include defining balanced multi-parameter objective functions and ensuring synthesizability. While ligand-based methods dominate, structure-based generative design is an emerging area, particularly valuable for orphan receptors and unexplored targets [14].

6. Optimizing Drug Candidates

Optimizing drug candidates is a crucial step in drug discovery, requiring careful balancing of efficacy, safety, pharmacokinetics, and manufacturability. AI has emerged as a transformative tool for enhancing this process by enabling rapid analysis of large datasets, predicting key molecular properties, and guiding decision-making. [15]

AI-driven virtual screening and lead optimization leverage deep learning and QSAR models to predict biological activity and prioritize potent, selective compounds [7].

ADMET prediction is enhanced through AI, reducing late-stage failures. Machine learning predicts solubility, toxicity, permeability, and metabolic stability [16].

Multi-objective optimization uses AI to evaluate potency, safety, and bioavailability simultaneously, tailoring candidates for clinical development and even specific patient populations [17].

Structure-based drug design is supported by tools like AlphaFold, which improve protein–ligand docking accuracy [18].

Reinforcement learning and retrosynthesis tools assess synthetic accessibility to ensure candidates are not just potent but also manufacturable [19].

7. Drug Repurposing

AI-driven drug repurposing is more cost-effective and faster than traditional methods, as it enables rapid screening of large compound libraries and reduces the need for extensive lab work. By leveraging high-throughput data, deep learning, and molecular modelling, AI identifies new therapeutic uses for existing drugs through side effect analysis, biomarker discovery, and mechanism elucidation.

Key databases and models support this process. Structural similarity, QSAR-based methods, and drug–response prediction platforms are central to modern AI-powered repurposing. These tools help uncover shared molecular pathways across diseases, improve precision medicine, and accelerate therapeutic strategy development [20].

How AI for Drug Discovery Monitoring Post-market safety:

Artificial intelligence (AI) has become a vital tool in the field of post-market drug safety monitoring. It makes it possible to continuously assess a drug's safety after it has received regulatory approval and been widely used by patients. Once a drug enters the market for AI-driven drug discovery, it becomes crucial for monitoring drug safety. This entails performing a number of crucial tasks [21]

Which include 

How AI ensures that the Medicines we use are Safe and Stay Safe

Figure 2: How AI ensures that the Medicines we use are Safe and Stay Safe

1)                  Signal Detection:  AI algorithms play a crucial role in analyzing extensive patient data archives. These algorithms enable the discovery of potential signals that indicate adverse events associated with specific drugs. This capability of AI drug discovery is instrumental in identifying uncommon or unforeseen side effects that may not have been evident during the initial clinical testing phases.[22]

2)                  Real-Time Monitoring: Real-time monitoring with AI is crucial in pharmacovigilance (drug safety surveillance). By processing data streams from healthcare databases and even social media posts, AI can uncover early warning signs of adverse drug reactions. This proactive system minimizes delays in response, improves patient safety, and supports regulatory authorities in making informed decisions. [22, 23]

3)                  Risk Prediction: Risk prediction in AI-driven drug discovery is a key aspect of precision medicine. By leveraging patient data such as genetics, medical history, and lifestyle, AI can forecast potential side effects before they occur. This proactive strategy not only reduces adverse drug reactions but also supports clinicians in choosing the safest and most effective therapy for each individual, leading to better clinical outcomes.[22,24,25]

4)                  Drug – Drug interaction: Drug–drug interaction prediction with AI is vital in clinical pharmacology and personalized medicine. Traditional detection often relies on post-market reports or clinical trials, which may not reveal all risks. AI models, however, can process vast datasets—including electronic health records, biochemical pathways, and pharmacological databases—to recognize hidden interaction patterns. This proactive approach reduces the likelihood of adverse events and supports safer prescribing practices.[23,24]

CURRENT TRENDS IN AI-DRIVEN DRUG DISCOVERY

1.                  Multimodal AI for Predicting Drug Combination Outcomes

A multimodal AI model that learns from structural, pathway, cell viability, and transcriptomic data to predict drug-combination effects across 953 clinical outcomes and 21,842 compounds, including combinations of approved drugs and novel compounds in development. Madrigal uses an attention bottleneck module to unify preclinical drug data modalities while handling missing data during training and inference, a major challenge in multimodal learning. It outperforms single-modality methods and state-of-the-art models in predicting adverse drug interactions, and ablations show both modality alignment and multimodality are necessary. It captures transporter-mediated interactions and aligns with head-to-head clinical trial differences for neutropenia, anaemia, alopecia, and hypo-glycemia. [26]

2.                  Generative AI & Protein-Structure Informed Drug Design

All the algorithm needs is a protein's three-dimensional surface structure. Based on that, it designs molecules that bind specifically to the protein according to the lock-and-key principle so they can interact with it.

Researchers have developed a new generative AI method that builds on decades of work in protein structure analysis and computer-aided drug design. Traditional approaches often relied on manual, time-consuming processes and frequently identified molecules that were difficult or impossible to synthesize. Until recently, AI was mainly used to refine existing compounds rather than create new ones.

The new system overcomes these limitations by automatically designing novel drug molecules that align precisely with a target protein’s 3D structure. It ensures the compounds are chemically synthesizable and selectively interact with the desired protein site while minimizing off-target effects—helping to reduce potential side effects. The AI was trained on extensive datasets of known chemical–protein interactions and structures.

In collaboration with Roche and other partners, the team successfully tested the method on PPAR proteins, which regulate sugar and fat metabolism. Since PPAR activators are already used in diabetes treatment to lower blood sugar, these results highlight the platform’s potential for developing effective, targeted therapies.  [27]

3.                  Integration of Multi-Omics & Systems Biology

Artificial intelligence (AI) in multiomics, a method that combines data from several "omic" approaches like transcriptomics, proteomics, genomics, and epigenomics, is having a big impact in the quickly changing field of healthcare. This creative combination combines the rich, diverse insights of multi-omics with the computational power of AI. According to Global Data, a top data and analytics company, it is transforming the domains of disease analysis and personalized medicine, signifying a substantial breakthrough in medical strategies and upcoming health interventions.[28]

4.                  AI in Clinical Trial Optimization

  Patient recruitment is a persistent challenge in clinical trials, necessitating automated, scalable solutions. TrialMatchAI is an AI-driven system designed to match patients to suitable trials by analyzing both structured and unstructured clinical data, including physician notes. Leveraging fine-tuned open-source large language models within a retrieval-augmented framework, the system normalizes biomedical entities, retrieves and ranks relevant trials, and evaluates eligibility at the criterion level using medical reasoning. This approach delivers transparent, explainable recommendations with traceable rationales.

  In real-world testing, TrialMatchAI successfully identified at least one relevant trial for 92% of oncology patients within the top 20 results. Its performance was validated across synthetic and real datasets, achieving over 90% accuracy in eligibility assessments, particularly for biomarker-driven matches. Designed for modularity, privacy, and lightweight deployment, the system supports standardized Phenopackets data, local installation, and easy integration of updated AI models. By improving efficiency, interpretability, and scalability, TrialMatchAI offers a robust solution for precision medicine-focused clinical trial matching. [29]

5.                  AI for Safety, Toxicology, and Off-Target Prediction

Artificial intelligence (AI) has become an essential tool in modern drug discovery, particularly for evaluating safety, predicting toxicity, and identifying off-target effects. In the area of drug safety, AI-powered systems process both structured and unstructured information—such as clinical records, pharmacovigilance reports, and adverse event databases—to uncover early warning signals and hidden drug–event relationships. For toxicology, advanced models like deep learning and in silicon screening platforms estimate cell- and organ-level toxic responses based on chemical structures, omics profiles, and high-throughput experimental data, helping reduce dependence on costly and lengthy animal studies. Off-target prediction benefits from AI methods including graph neural networks, similarity-based modelling, and structure-informed learning, which can map unintended drug–protein interactions, anticipate harmful effects, and highlight opportunities for drug repurposing. Collectively, these approaches speed up decision-making, minimize late-stage failures in clinical development, and improve overall patient safety. Looking forward, emerging directions include the adoption of explainable AI, patient-specific digital twin models for personalized safety evaluation, and broader regulatory acceptance of AI-based toxicology frameworks. [32]

6.                  Democratization & Platformization of AI Tools

Democratization and Platformization of AI tools are making advanced technologies widely accessible beyond large pharmaceutical companies. User-friendly platforms and open-source frameworks allow non-experts to apply AI in tasks like virtual screening, safety evaluation, and toxicity prediction. Integrated platforms further streamline workflows by combining data management, modelling, and visualization in one system. Together, these trends lower barriers to entry, promote collaboration, and accelerate innovation in drug discovery. [33, 34]

7.                  Real-World & Prospective Validation is Rising

AI applications in drug discovery are increasingly moving from retrospective analyses to real-world and prospective validation. Models are now tested on real-world data, including electronic health records, registries, and post-marketing surveillance, to assess their reliability and clinical utility. Prospective studies, where AI predictions are evaluated in ongoing experiments or clinical trials, are becoming more common, bridging the gap between theoretical accuracy and practical applicability. This shift enhances confidence in AI systems, reduces bias, and supports regulatory adoption in pharmaceutical development. [7, 35, 36, 37]

8.                  Regulatory & Ethical Trends

Regulatory and ethical considerations are becoming central to AI integration in drug discovery. Agencies like the FDA and EMA are establishing guidelines to ensure AI models are validated, reproducible, and explainable in preclinical and clinical applications. Ethical concerns, including data privacy, bias mitigation, and equitable access, are also being addressed to prevent harm or discrimination. Together, these trends promote the development of AI systems that are both trustworthy and compliant with safety and societal standards. [43, 44, 45]

Regulatory Trends

·                     FDA’s Role in AI-Driven Drug Development
The FDA recognizes the growing integration of AI across all stages of drug development, including nonclinical research, clinical trials, manufacturing, and post marketing monitoring. The agency is actively developing strategies to provide clear regulatory guidance for AI and machine learning (AI/ML) applications, taking into account both the opportunities and challenges associated with their use in the drug development process.[43]

·                     Regulating AI in Drug Development: Legal Challenges and Compliance Strategies
This work examines the regulatory and legal challenges surrounding AI in drug development and outlines compliance strategies. Key considerations include ensuring high-quality, representative data to reduce algorithmic bias, addressing the “black box” nature of many AI models, and maintaining continuous validation to manage AI’s dynamic nature throughout its lifecycle. [44]

Ethical Considerations

Ethical and Bias Considerations in Artificial Intelligence
The article discusses the ethical issues and bias challenges in AI systems, stressing the need to mitigate algorithmic bias and promote fairness. It highlights the FAIR principles as a framework to guide ethical AI development by minimizing biases from the earliest stages of data management.[45]

Ethical Considerations in AI-Driven Drug Discovery
This study focuses on ethical concerns specific to AI in drug discovery, including data privacy, algorithmic bias, and transparency. It underscores the importance of designing AI systems that protect patient privacy, ensure equitable outcomes, and actively reduce biases in both data and algorithms.[45]

9.                  Market Growth & Investments

·         Global Market Expansion: The AI in drug discovery market was valued at USD 1.9 billion in 2024 and is projected to reach USD 9.1 billion by 2030, growing at a CAGR of 29.7% .[46]

·         Long-Term Forecast: By 2034, the market is expected to surpass USD 16.5 billion, driven by advancements in AI technologies and increasing adoption across the pharmaceutical industry. [47]

·         Regional Dynamics: North America currently holds a significant market share, while the Asia-Pacific region is anticipated to experience the fastest growth, with a projected CAGR of 21.1% from 2025 to 2034 .[47]

Investment Trends & Funding Highlights

·         Generative AI Focus: Generative AI in drug discovery is gaining momentum, with the market expected to grow from USD 318.55 million in 2025 to over USD 2.8 billion by 2034, reflecting a CAGR of 27.42%. [48]

·         Venture Capital Activity: In 2024, equity funding for AI in drug discovery reached USD 3.8 billion, up from USD 3 billion in 2023, indicating a growing investor confidence in AI-driven pharmaceutical innovations .[49]

Key Drivers of Market Growth

·         Cost Reduction & Efficiency: AI technologies are enabling pharmaceutical companies to reduce the time and cost associated with drug discovery, making the process more efficient and cost-effective.

·         Advancements in AI Capabilities: Continuous improvements in AI algorithms and computational power are enhancing the ability to predict molecular properties and optimize drug designs.

·         Strategic Collaborations: Partnerships between AI firms and pharmaceutical companies are accelerating the development and application of AI in drug discovery, fostering innovation and expanding market opportunities.[50]

CHALLENGES OF AI IN DRUG DISCOVERY

While AI shows great promise in transforming pharmaceutical research, its broad adoption faces significant hurdles. Challenges include technical constraints, limited or fragmented data, unclear regulatory frameworks, and ethical considerations. Overcoming these obstacles is crucial for AI to evolve from experimental applications into trusted tools within mainstream drug development. [51]

Figure 3: Challenges of AI in Drug Discovery

1.                  Data Quality, Availability, and Standardization

Reliable AI in drug discovery depends on high-quality, accessible, and well-standardized data. Models require large, accurate, and diverse datasets—such as chemical structures, omics profiles, clinical records, and high-throughput screening results—to deliver dependable predictions. However, data is often fragmented, inconsistent, or isolated across institutions, limiting model accuracy and reproducibility. Implementing standardized formats, ontologies, and FAIR (Findable, Accessible, Interoperable, Reusable) principles is critical for integrating heterogeneous datasets and enabling AI to generate robust, generalizable insights. [51, 4]

1.                     Model Interpretability and Transparency

A major challenge in AI-driven drug discovery is the lack of interpretability in deep learning models, which act as “black boxes” and offer little explanation for their predictions. This limits trust and regulatory acceptance. Explainable AI (XAI) methods—such as SHAP, LIME, and attention mechanisms—aim to clarify how models make decisions by highlighting influential molecular or biological features. However, their adoption remains limited, and balancing interpretability with predictive accuracy continues to be difficult. [52]

2.                  Validation and Reproducibility

When using AI for drug discovery, reproducibility and validation are major obstacles. When applied to independent or real-world data, many AI models lose accuracy despite performing well on benchmark datasets. Inadequate external validation, data heterogeneity, and overfitting are frequently the causes of this lack of reproducibility. It can also be challenging to compare studies or replicate findings due to discrepancies in data pre-processing, feature selection, and evaluation metrics.

The need for strong validation pipelines, which include independent replication studies, cross-cohort testing, and prospective validation, is becoming more and more stressed by regulatory bodies and the scientific community. To guarantee the reproducibility of AI-driven discoveries, standardization of datasets, reporting procedures, and benchmarking frameworks is essential. The translational potential of AI, from clinical applications to computational predictions, is still constrained in the absence of such rigor. [53]

3.                  Regulatory, Ethical, and Legal Issues

The application of AI in drug discovery introduces significant regulatory, ethical, and legal challenges. Current regulatory systems, such as those established by the FDA and EMA, were built for conventional drug development models and may not adequately capture the unique complexities of AI-based methodologies. Major concerns include establishing clear validation standards for AI tools, ensuring adherence to Good Machine Learning Practices (GMLP), and creating defined pathways for regulatory approval.

From an ethical perspective, issues surrounding data privacy, informed consent, and algorithmic bias remain pressing. AI models developed on incomplete or biased datasets risk producing unfair or unequal outcomes, which may disadvantage specific patient groups. Additionally, accountability remains ambiguous—particularly regarding responsibility for errors or misjudgements generated by AI systems. Intellectual property (IP) rights further complicate the landscape, especially with respect to the ownership of AI-designed molecular entities.

To overcome these hurdles, collaboration across multiple stakeholders—including regulators, AI scientists, pharmaceutical industries, and ethicists—is essential. The development of transparent governance policies, globally harmonized AI regulations, and clear ethical frameworks will be crucial to ensure equity, trust, and compliance in AI-driven drug discovery. [54]

4.                  Technical and Computational Barriers

Despite its potential, the use of AI in drug discovery faces significant technical and computational limitations. Training advanced AI models, particularly deep learning and generative models, requires vast amounts of high-quality data and substantial computational power. Access to such resources is often restricted to large pharmaceutical companies or well-funded institutions, creating a gap for smaller research groups. Moreover, managing and processing massive datasets—such as genomic, proteomic, and chemical libraries—poses additional challenges in terms of storage, scalability, and integration across platforms.

Another major issue lies in the complexity of biological systems, which makes it difficult for AI models to fully capture dynamic interactions at the molecular, cellular, and systemic levels. Simplifications or incomplete modelling may lead to inaccurate predictions. Additionally, interoperability problems between different software tools, lack of standardized data formats, and limited model generalizability across diverse therapeutic domains further hinder progress.

Overcoming these barriers will require advances in computational infrastructure, more efficient algorithms, cloud-based platforms, and collaborative data-sharing frameworks. Investments in scalable technologies and open-access resources could democratize the benefits of AI in drug discovery and accelerate innovation. [3]

5.                  Cultural and Organizational Challenges

Beyond technical limitations, the adoption of AI in drug discovery is also hindered by cultural and organizational barriers. Many pharmaceutical companies remain cautious about integrating AI into their workflows due to skepticism, lack of trust in model predictions, and limited understanding of advanced computational methods among traditional researchers. Resistance to change is common in established organizations where conventional drug development practices are deeply entrenched.

Another challenge is the skills gap there is a shortage of professionals who can bridge expertise in both life sciences and AI technologies. This limits effective collaboration between computational scientists and biomedical researchers. Additionally, organizational silos, fragmented communication, and competition between departments can slow down AI adoption. For smaller companies and academic labs, financial constraints and uncertainty about the return on investment in AI-driven approaches also create reluctance.

Addressing these issues requires cultural transformation within the pharmaceutical ecosystem, emphasizing cross-disciplinary training, workforce upskilling, and collaborative environments where data scientists and domain experts can work together. Leadership support, open communication, and change management strategies are crucial to foster trust in AI and encourage its seamless integration into drug discovery pipelines. [6]

FUTURE PROSPECTS OF AI IN DRUG DISCOVERY

The impact of artificial intelligence on advancements in pharmacology and the pharmaceutical industry.

Drug discovery is traditionally a lengthy, expensive, and inefficient process, often requiring over a decade of research and billions of dollars in investment, yet with relatively low success rates. Artificial intelligence (AI) offers a transformative solution by enabling the rapid analysis of large biomedical datasets, predicting drug–target interactions, assessing ADMET profiles, and optimizing clinical trial design. These capabilities can reduce costs, improve precision, and accelerate the development of new therapeutics.

Looking ahead, the integration of AI with multi-omics technologies, real-world evidence, and digital twin models will enable highly personalized medicine, while advances in generative AI and quantum computing are expected to revolutionize de novo drug design and molecular simulations. Although still in its formative stage, AI-driven drug discovery is poised to reshape pharmaceutical R&D, shorten development timelines, and expand therapeutic innovation across diverse disease areas. [7, 50]

Figure 4: Revolutionizing drug discovery:

1.                  Personalized Medicines

Personalized medicine aims to tailor medical treatment to the individual characteristics of each patient, considering genetic, molecular, and environmental factors. AI plays a pivotal role in enabling this approach by integrating multi-omics data—including genomics, transcriptomics, proteomics, and metabolomics—with clinical information from electronic health records, wearable devices, and patient registries. By analyzing these large and complex datasets, AI can identify patient-specific therapeutic targets, predict drug responses, and stratify patients for optimized treatment plans.

Digital twin technologies further enhance personalized medicine by creating virtual models of individual patients. These models simulate responses to different drug candidates, allowing researchers to predict efficacy and safety before actual administration. AI-driven predictive modelling also enables biomarker discovery, drug repurposing, and optimization of dosing regimens, thereby reducing adverse effects and improving clinical outcomes.

The integration of AI into personalized medicine promises not only to accelerate drug development but also to make treatments more precise, efficient, and patient-centered. By combining computational power with biological and clinical insights, AI facilitates the transition from a “one-size-fits-all” approach to truly individualized therapeutic strategies. [55, 56]

2.                  Accelerated Drug Development

Traditional drug development is notoriously slow, often taking 10–15 years from discovery to market approval, with high attrition rates in preclinical and clinical stages. AI has the potential to significantly accelerate this process by improving target identification, optimizing lead compounds, predicting ADMET (absorption, distribution, metabolism, excretion, and toxicity) profiles, and streamlining clinical trial design. By analysing large-scale biomedical datasets and simulating molecular interactions, AI reduces reliance on trial-and-error experimentation, enabling faster progression from discovery to clinical testing.

Advanced AI techniques, including generative models, predictive algorithms, and digital twins, facilitate the design of molecules with desired pharmacological properties while anticipating potential side effects. Integration with high-throughput screening platforms and automated synthesis systems further expedites compound testing and optimization. Collectively, these innovations promise to shorten development timelines, reduce costs, and improve the overall success rate of new therapeutics. [7, 57, 50]

3.                  Improved Success Rate

One of the major challenges in traditional drug development is the low success rate, with most drug candidates failing during preclinical or clinical stages. AI has the potential to enhance success rates by accurately predicting drug–target interactions, toxicity, pharmacokinetic properties, and off-target effects early in the development process. Machine learning algorithms can analyze large and diverse datasets to identify high-potential compounds, reducing the likelihood of failure in later stages.

By integrating predictive modelling, multi-omics data, and real-world evidence, AI can also improve patient stratification for clinical trials, ensuring that therapies are tested on populations most likely to respond. Generative AI models further allow the design of optimized molecules with favourable ADMET profiles, reducing attrition due to pharmacological or safety issues. Collectively, these strategies increase the probability that drug candidates advance successfully from discovery to market approval. [7, 50, 51]

4.                  Data Driven Insights

Data-driven insights form the foundation of AI’s transformative impact on drug discovery. The pharmaceutical sector produces vast and diverse datasets, including genomic and proteomic information, chemical compound libraries, clinical trial outcomes, and real-world patient records. AI and machine learning techniques can efficiently process and analyze these complex datasets to uncover patterns, predict drug–target interactions, and identify novel therapeutic candidates far more effectively than traditional approaches.

By combining information from multiple sources, AI systems provide actionable insights for tasks such as target identification, lead optimization, and patient stratification. Predictive models also allow for the early detection of potential toxicity or off-target effects, minimizing late-stage failures. In addition, AI-driven analytics support biomarker discovery, drug repurposing, and personalized therapy design, enhancing the precision, cost-effectiveness, and patient-centricity of drug development. [5, 8, 37, 58]

5.                  Enhanced Clinical Trials

AI is playing an increasingly important role in the planning, execution, and evaluation of clinical trials, enhancing their efficiency, adaptability, and patient focus. By leveraging large-scale datasets from electronic health records, wearable technologies, and prior study outcomes, AI can identify appropriate patient populations, refine inclusion and exclusion criteria, and forecast individual treatment responses. These capabilities improve patient recruitment, lower dropout rates, and increase the probability of trial success.

Additionally, AI facilitates adaptive clinical trials, allowing modifications to study parameters in real time based on interim findings. Predictive models can anticipate adverse events and optimize dosing, improving both safety and trial efficiency. AI-driven analytics further support biomarker identification, patient stratification, and outcome prediction, helping reduce costs and timelines while enhancing the reliability and quality of clinical trial results. [7, 50, 58]

6.                  Novel Therapies

AI is accelerating the development of novel therapies by identifying unexplored molecular targets and designing new compounds with optimized pharmacological properties. Generative AI enables de novo drug design, while integration with patient-specific genomic and proteomic data facilitates personalized and precision medicine. Additionally, AI-driven drug repurposing uncovers new therapeutic uses for existing drugs, offering cost-effective solutions for rare and neglected diseases. [7, 50, 58]

7.                  Integration of AI with Multi-Omics and Systems Biology

Integrating AI with multi-omics technologies—including genomics, transcriptomics, proteomics, metabolomics, and epigenomics—provides transformative opportunities in drug discovery AND development. AI algorithms can process and interpret complex, high-dimensional datasets to uncover new therapeutic targets, biomarkers, and molecular pathways, offering a comprehensive, systems-level understanding of disease mechanisms. This approach also enables predictive modeling of drug responses and patient stratification, supporting the advancement of precision medicine.

When combined with systems biology frameworks, AI can simulate complex biological networks, model interactions among genes, proteins, and metabolites, and predict the effects of interventions, such as drug treatments, on cellular and organismal functions. These insights facilitate rational drug design, optimization of combination therapies, and identification of potential off-target effects, ultimately improving the efficiency and success rates of drug development. Therefore, the integration of AI with multi-omics and systems biology provides a holistic and strategic approach to understanding disease and designing targeted therapies. [59, 60]

8.                   Regulatory Evolution and Standardization

The integration of AI into drug discovery necessitates adaptation and evolution of regulatory frameworks. Traditional regulations established by agencies such as the FDA and EMA were designed for conventional drug development pipelines and may not fully address the complexities of AI-driven approaches. Regulators are increasingly exploring guidelines for validation, transparency, reproducibility, and Good Machine Learning Practices (GMLP) to ensure that AI-generated predictions are reliable and actionable in clinical settings.

Standardization is crucial for fostering trust and interoperability. Consistent data formats, reporting standards, and evaluation metrics enable regulatory agencies, pharmaceutical companies, and research institutions to assess AI models’ performance and ensure patient safety. International collaboration harmonization of guidelines are becoming essential to address cross-border challenges in AI-enabled drug development. Clear regulatory pathways will not only facilitate the adoption of AI technologies but also ensure ethical, safe, and reproducible practices across the industry. [60]

CONCLUSION:

Artificial intelligence is rapidly transforming drug discovery by addressing key challenges such as long development timelines, high costs, and low success rates. Its current applications—from virtual screening to clinical trial optimization—demonstrate its versatility and potential to disrupt traditional processes. However, issues like data quality, model transparency, regulatory uncertainty, and high computational demands remain significant hurdles.

Looking ahead, integrating AI with multi-omics, systems biology, and digital health could pave the way for personalized medicine and faster development of new therapies. Techniques like federated learning, explainable AI, and human–AI collaboration may enhance trust and support regulatory approval. Additionally, combining AI with emerging technologies like quantum computing and synthetic biology could further accelerate innovation.

Ultimately, AI is driving a shift from trial-and-error approaches to data-driven drug development. Overcoming current challenges will require collaboration across academia, industry, and regulatory bodies. If fully leveraged, AI could make drug discovery faster, more affordable, and more patient-centred.

REFERENCES:

1.      Russell S, Norvig P. Artificial Intelligence: A Modern Approach. 4^th ed. Pearson Education; 2021.

2.      Kaplan A, Haenlein M. Siri, Siri, in my hand: Who’s the fairest in the land? On the interpretations, illustrations, and implications of artificial intelligence. Business Horizons. 2019; 62(1):15–25. Doi:10.1016/j.bushor.2018.08.00

3.      Mak KK, Pichika MR. Artificial intelligence in drug development: Present status and future prospects. Drug Discovery Today. 2019; 24(3):773–780. doi:10.1016/j.drudis.2018.11.014

4.      Paul SM, Mytelka DS, Dunwiddie CT, Persinger CC, Munos BH, Lindborg SR, Schacht AL. How to improve R&D productivity: The pharmaceutical industry's grand challenge. Nat Rev Drug Discovery. 2010; 9(3):203–214. doi:10.1038/nrd3078

5.      Schneider G, Clark DE. Automated de novo drug design: Are we nearly there yet?                                                                  Angew Chem Int Ed Engl. 2019; 58(32):10792–10803. doi:10.1002/anie.201814843

6.      Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, et al. Highly accurate protein structure prediction with alphaFold. Nature. 2021; 596(7873):583–589. doi:10.1038/s41586-021-03819-2

7.      Vamathevan, J., Clark, D., Czodrowski, P. et al. (2019). Applications of machine learning in drug discovery and development. Nature Reviews Drug Discovery, 18(6), 463–477. https://doi.org/10.1038/s41573-019-0024-5

8.      McNair D, Young M, Annetta M. Artificial intelligence and machine learning for lead-to-candidate decision-making and beyond. Annu Rev Pharmacol Toxicol. 2022;62:xxx–xxx. Doi:10.1146/annurev-pharmtox-051921-023255.

9.      Subramanian, A., Narayan, R., Corsello, S. M., Peck, D. D., Natoli, T. E., Lu, X., ... & Golub, T. R. (2017). A next generation connectivity map: L1000 platform and the first 1,000,000 profiles. Cell, 171(6), 1437–1452.e17. https://doi.org/10.1016/j.cell.2017.10.049

10.  Freshour, S. L., Kiwala, S., Cotto, K. C., Coffman, A. C., McMichael, J. F., Song, J. J., ... & Griffith, M. (2021). Integration of the Drug–Gene Interaction Database (DGIdb 4.0) with open crowdsource efforts. Nucleic Acids Research, 49(D1), D1144–D1151. https://doi.org/10.1093/nar/gkaa1084

11.  Priyakumar UD, Deshpande S, Joshi R, Sonavane U, Dandekar P. Applications of machine learning in computer-aided drug discovery. WIREs Comput Mol Sci. 2022;12(6):e1581. doi:10.1002/wcms.1581

12.  Ramsundar B, Liu B, Wu Z, Verras A, Tudor M, Feinberg EN, et al. Deep Learning for the Life Sciences: Applying Deep Learning to Genomics, Microscopy, Drug Discovery, and More. 1st ed. Sebastopol (CA): O’Reilly Media; 2019.

13.  Pillai, N., Abos, A., Teutonico, D., & Mavroudis, P. D. M. (2024). Machine learning framework to predict pharmacokinetic profile of small molecule drugs based on chemical structure. Clinical and Translational Science

14.  Brown, N., Fiscato, M., Segler, M. H. S., & Vaucher, A. C. (2019).

15.  GuacaMol: Benchmarking Models for de Novo Molecular Design.

16.  Journal of Chemical Information and Modeling, 59(3), 1096–1108.

17.  Xia, Y., Wang, Y., & Zhang, W. (2024). A comprehensive review of molecular optimization in artificial intelligence based drug discovery. Quantitative Biology, 12(1), 15-42.https://doi.org/10.1002/qub2.30

18.  Ekins, S. et al. "Machine learning models and pathway prediction for ADMET." Nature Reviews Drug Discovery, 18(11), 696–714 (2019).

19.  DOI: 10.1038/s41573-019-0024-5

20.  Segler, M.H.S., Preuss, M., and Waller, M.P. "Planning chemical syntheses with deep neural networks and symbolic AI." Nature, 555, 604–610 (2018).DOI: 10.1038/nature25978

21.  Jumper, J. et al. "Highly accurate protein structure prediction with AlphaFold." Nature, 596, 583–589 (2021).DOI: 10.1038/s41586-021-03819-2

22.  Coley, C.W. et al. "Machine learning in computer-aided synthesis planning." Accounts of Chemical Research, 51(5), 1281–1289 (2018). DOI: 10.1021/acs.accounts.8b00075

23.  Pushpakom, S., Iorio, F., Eyers, P.A., Escott, K.J., Hopper, S., Wells, A., Doig, A., Guilliams, T., Latimer, J., McNamee, C., et al. "Drug repurposing: progress, challenges and recommendations." Nature Reviews Drug Discovery, 18(1), 41–58 (2019). DOI: 10.1038/nrd.2018.168

24.  Smith, J., & Lee, A. (2022). The role of artificial intelligence in post-market drug safety monitoring. Journal of Pharmacovigilance, 15(3), 145-160.

25.  Nagar, A , Gobbiru, J., & Chakravarty, A.(2025) Artificial Intelligence in Pharmacovigilance: advancing drug safety monitoring & regulatory integration.

26.  Wang , L., Huang , Y. Chen, J., & Zhang, W.(2024) AI & big data for Pharmacovigilance and Patient safety, Journal of Medicine , Surgery and Public health. https://appinventiv.com/blog/ai-in-drug-discovery /

27.  Algarvio , R. C., Almeida, A I., Gomes , J. J., et al (2025). AI in Pharmacovigilance : a narrative review and practical experience with an expert defined Bayesian network  tool . International journal of Clinical Pharmacy .

28.  Yepeng Huang, Xiaorui Su, Varun Ullanat, Intae Moon, Ivy Liang, Lindsay Clegg, Damilola Olabode, Ruthie Johnson, Nicholas Ho, Megan Gibbs, Alexander Gusev, Bino John & Marinka Zitnik. Multimodal AI predicts clinical outcomes of drug combinations from preclinical data

29.  Atz, K., Cotos, L., Isert, C., Håkansson, M., Focht, D., Hilleke, M., Nippa, D. F., Iff, M., Ledergerber, J., Schiebroek, C. C. G., Romeo, V., Hiss, J. A., Merk, D., Schneider, P., Kuhn, B., Grether, U., & Schneider, G. (2024). Prospective de novo drug design with deep interactome learning.

30.  “AI integration with multiomics: Transformative leap in healthcare, says GlobalData.” GlobalData, Disruptor Intelligence Center, “AI integration with multiomics: Transformative leap in healthcare, says GlobalData”, 24 Jan 2024.

31.  Abdallah, M., Nakken, S., Bierkens, M., Galvis, J., Groppi, A., Karkar, S., et al. (2025). TrialMatchAI: An End-to-End AI-powered Clinical Trial Recommendation System to Streamline Patient-to-Trial Matching

32.  Zhou, X., Yang, C., Liu, Z., Li, S., Chen, C., & Yu, H. (2024). LLM-Match: An Open-Sourced Patient Matching Model Based on Large Language Models and Retrieval-Augmented Generation.

33.  Patel, Y. R., Li, R. C., Ngo, J., et al. (2023). TrialGPT: Matching patients to clinical trials with large language models. NPJ Digital Medicine, 6, Article 98. https://doi.org/10.1038/s41746-023-00880-2

34.  Tran, T. T. V., Wibowo, A. S., Tayara, H., & Chong, K. T. (2023).
Artificial Intelligence in Drug Toxicity Prediction: Recent Advances, Challenges, and Future Perspectives. Journal of Chemical Information and Modeling, 63(9), 2628‑2643.

35.  The Role of AI in Drug Discovery — Abbas, 2024, ChemBioChem

36.  Open‑Source Browser‑Based Tools for Structure‑Based Computer‑Aided Drug Discovery — Wang & Durrant, 2022, Molecules

37.  Beam, A. L., & Kohane, I. S. (2018). Big Data and Machine Learning in Health Care. JAMA, 319(13), 1317–1318. DOI: 10.1001/jama.2017.18391

38.  Wiens, J., Saria, S., Sendak, M., et al. (2019). Do no harm: a roadmap for responsible machine learning for health care. Nature Medicine, 25, 1337–1340. DOI: 10.1038/s41591-019-0548-6

39.  U.S. Food and Drug Administration (FDA). (2021). Artificial Intelligence and Machine Learning in Software as a Medical Device.