




































Agrawal and Kulkarni                                                                          AI Applications in Drug Discovery and Development 

Online First                                                                                                                         Indian J Pharm Drug Studies | 1  

Review Article 

Application of Artificial Intelligence in drug discovery, designing, clinical trials 

and repurposing 

Rashmi Agrawal1, Meghashyama Kulkarni2 

From, 1Consultant, Devmata Hospital, Bhopal, Madhya Pradesh, India, 2Consulting Oral Pathologist, Bengaluru, Karnataka, India 

ABSTRACT  

Drug development, right from discovery, designing, and repurposing, along with various phases to clinical trials, is a complex and time-

consuming process that traditionally relies on the experience of labor forces, along with incessant trial-and-error experimentation. The 

emergence of various artificial intelligence (AI) tools and techniques is redefining the pharmaceutical industry. The integration of AI-

driven methodologies into all stages of the drug development pipeline has enhanced the efficiency and effectiveness of the process. 

With this review, we aim to provide insights into the application of AI at various stages mentioned, with an emphasis on the latest 

advancements. The role of AI in speeding up the process of drug development while giving distinct and accurate outcomes is highlighted 

as well. Finally, we addressed the current challenges in employing AI along with future perspectives in order to enhance AI-augmented 

drug development, ultimately offering significant benefits to patients and society. 

Key words: Artificial intelligence, drug discovery, clinical trial design, neural networks, drug repurposing

rtificial Intelligence (AI) involves the least amount of 

human intervention possible when utilizing a 

computer to imitate intelligent behaviour [1]. AI has 

been implemented in the pharmaceutical industry in recent 

years. One of the important aspects in medicinal chemistry, 

drug discovery that involves identifying, formulating, and 

developing new medications, is an arduous, technique-

sensitive, and time-consuming process with long-established 

protocols that require exorbitant resources with several trial and 

error experimentations needing a good deal of time [2]. This 

complex process, further requires an average of 12 years and a 

cost of $2.6 billion to forge ahead a single molecule from 

inception till Food and Drug Administration (FDA) approval. 

In spite of these strenuous endeavours, the process is marked 

by unfavourable adverse effects of the drugs, high attrition 

rates, and tenacious challenges in addressing chronic diseases, 

including cancers and diabetes mellitus [3, 4].  

The emergence of AI marks a revolutionary drift in drug 

development, with the advent of computational tools designed 

to supplement human capabilities and intelligence rather than 

replace them [5]. AI leverages suave algorithms for 

autonomous decision-making, hence revolutionizing the 

pharmaceutical industry [6]. AI techniques such as machine 

learning (ML) and deep learning (DL) offer the potential to 

accelerate the above-mentioned processes by enabling more 

coherent, efficient, and precise analysis of large amounts of 

data [7]. AI has also been able to predict the toxicity of drugs;  

Access this article online 

 

Received –  23rd July 2025 

Initial Review –  24th July 2025 

Accepted – 25th July 2025 

Quick Response Code 

besides its role in drug-target interaction predictions, clinical 

trial design, and drug repurposing, it is aiding in the reformation 

of drug development approaches [8, 9]. 

This narrative review aims to unveil the transformative 

impact of AI on the whole process of drug discovery till the 

completion of development and sheds light on the role of AI in 

drug repurposing as well. The paper also emphasizes the 

importance of AI in accelerating the development of novel 

therapeutics while juxtaposing it with traditional 

methodologies. Furthermore, we also discuss the ongoing 

challenges with respect to the usage of AI, along with the future 

of AI in the pharmaceutical industry and the ways it can 

significantly restructure the global healthcare system. 

Revolution of AI  

The Dartmouth Workshop conducted in 1956 is considered the 

foundational event for the advent of AI. It is described as 

intelligence displayed by man-made machines, dedicated to 

developing theories, technologies, applications, and methods 

aimed at replicating, enhancing, and extending human 

intelligence [10]. Over the past 70 years, AI has evolved from 

a theoretical notion into a powerful industrial adaptation, 

revolutionizing industries such as manufacturing, agriculture, 

finance, and healthcare [11-13]. 

AI includes various subtypes, some of which are described 

as follows: machine learning (ML), Natural Language 

Processing (NLP), Computer Vision (CV), and Fuzzy Logic  

__________________________________________________ 

Correspondence to: Dr. Rashmi Agrawal. Devmata Hospital, 

Bhopal, Madhya Pradesh, India. 

Email: atharvapub@gmail.com 

A 

mailto:atharvapub@gmail.com


Agrawal and Kulkarni                                                                          AI Applications in Drug Discovery and Development 

Online First                                                                                                                         Indian J Pharm Drug Studies | 2  

(FL). ML works on algorithms trained for decision-making that 

learn from the already analyzed data. ML is further classified 

into supervised learning and unsupervised learning. In 

supervised learning, pre-catalogued/ labelled data is used as 

input, whereas in the latter, training data is not 

catalogued/labelled, and the system must perceive and label the 

said data. A third sub-category exists, known as semi-

supervised learning, a combination of the two previously 

explained types [14]. Deep learning (DL) is a subgroup of ML 

based on systems that use artificial neural networks (ANN), 

which imitate the human brain, and act by interpreting and 

drawing conclusions from the given data. Whereas, NLP is a 

branch of AI that recognizes natural language and builds 

communication between machines and humans. CV is a 

subtype of AI that allows computers to discern an image and 

distinguish the individual elements of that particular image by 

assigning them a meaning. On the other hand, FL uses non-

binary values to solve issues that normally require tackling with 

more values, hence deciphers the problems that classical logic 

cannot solve [15, 16]. 

 

Figure 1: Classification of ANNs [16, 17] 

An ML algorithm for any task begins with task definition 

and ends with model application. Different steps with detailed 

explanations are described in Table 1. 

Table 1: Steps in Machine Learning [16, 17] 

S. no Step Description 

1 Pre-processing  Ensures that the algorithm easily 
interprets the datasets 

 Includes data loading, normalization, 
aggregation, and standardization of 
the dataset 

2 Exploratory 

Data Analysis 

(EDA) 

 Tests several hypotheses and yields a 
better understanding of dataset 

variables and their inter-
relationships. 

 Check for errors and missing values 

 Helps determine if the statistical 
techniques that are used for the data 

analysis are appropriate. 

3 Model Selection  Creation of a predictive model 

 Algorithms are often grouped 
according to the ML techniques used 
such as supervised or unsupervised 
etc.  

 Few ML algorithms used in medical 
research are Support Vector Machine 
(SVM), Multinomial Logistic 
Regression (MLR), Bayesian 

networks (BN), and Decision Tree 
(DT). 

 Few DL algorithms used in medical 
research are all the sub-types of 
ANNs, Long short-term memory 
(LSTM), and mixed networks. 

4 Model 

Processing and 

Evaluation  

 Datasets are divided into test and 
train sets, and further cross-
validation is done. 

 If the model shows discrepancies, it 
can be rebuilt using improvement 
strategies, which is known as 
‘TUNING’ 

 Model evaluation is done with 
ANNs. 

Data, computation, and algorithms, the essential components of 

AI, serve as the basis of the AI-driven pharmaceutical research 

[18]. Sources of data in the pharmaceutical industry include 

public and commercial datasets, and research datasets 

generated through data mining. The evolution of computational 

power and the advent of various algorithms have provided 

critical support for AI pharmaceutical companies in carrying 

out research using the extremely potent tool, AI [19]. 

Challenges faced with traditional drug discovery and 

development methods 

Traditional drug discovery, as described before, is a complex, 

time-consuming, laborious, and costly undertaking. To bring a 

single drug to the market, it typically takes over a decade and 

an average cost exceeding $2 billion. Each and every stage, 

starting from target identification and validation till preclinical 

testing and clinical trials, requires immense trial-and-error 

experimentations with huge irreversible expenses along the 

way [20]. Furthermore, attrition rates of the drug candidates are 

very high, with almost 90% of them failing owing to 

insufficient efficacy and/or safety concerns during clinical 

trials. Additionally, financial losses multiply when toxicity is 

detected after-market release of the drugs [21]. 

Hence, evaluation of toxicity and safety is of paramount 

importance during drug development, which can be achieved 

by a better understanding of the protein–ligand interactions and 

predictability of toxicity. The development of computational 

methods and AI has emerged as a promising approach to tackle 

the above-mentioned obstacles while developing a drug. The 

further sections delve into the applications of AI in all the steps 

of drug development [21]. 

Role of AI in drug discovery  

There exist more than 1060 molecules that help in the 

development of novel drugs [22]. Al has the ability to collect 

https://pubmed.ncbi.nlm.nih.gov/38758610
https://pubmed.ncbi.nlm.nih.gov/38758610


Agrawal and Kulkarni                                                                          AI Applications in Drug Discovery and Development 

Online First                                                                                                                         Indian J Pharm Drug Studies | 3  

positional information about molecules within the space to 

search for bioactive compounds and select appropriate 

molecules via virtual screening (VS). A few chemical spaces 

that have open access are ChemBank, PubChem, ChemDB, and 

DrugBank [17]. Quantitative structure-activity relationship 

(QSAR)-based computational models are capable of quickly 

predicting large numbers of physicochemical parameters of the 

drug candidates. AI-based QSAR models, such as support 

vector machines (SVMs), linear discriminant analysis (LDA), 

random forest (RF), and decision trees, are being applied to 

speed up QSAR analysis [23]. 

Deep Docking (DD), an open-source protocol for AI-

enabled VS, was developed by Gentile and colleagues. It is 

supposed to be one of the fastest AI-enabled docking platforms 

being tested on one billion-plus molecular libraries [24]. The 

input data in DD consists of a particular molecule’s Simplified 

Molecular Input Line Entry System (SMILES) and the target’s 

structure. DD executes molecular docking for a small subset of 

a large library to deduce the ranking of the unprocessed 

datasets, followed by ligand-based prediction for the rest of the 

library. In this way, DD scraps off undockable molecular 

structures without wasting computational resources [25]. 

Further, diverse ML techniques, including Naïve Bayesian 

algorithm (NB),  k-nearest neighbor algorithm (kNNs), support 

vector machine (SVMs), and ANNs, can also be used for VS. 

Although ANNs and SVMs are commonly considered as the 

most accurate, NB is excellent in identifying favourable 

scaffold fragments, while kNN is easy to implement and utilizes 

multi-task learning (MTL) {MTL training, a sub-field of ML, 

involves training a model to perform multiple tasks at the same 

time}. Amalgamating various ML algorithms is preferred as it 

can enhance performance [26]. 

DL algorithms are proven to show increased predictability 

compared to traditional ML models, especially in distribution, 

metabolism, excretion, and toxicity (ADMET) data sets [27]. 

The recent rise of Graph Neural Networks (GNNs) has set in 

motion a new paradigm in ADMET model designing. GNNs 

offer an informative and compact representation of datasets 

[28]. The efficacy of GNN algorithms in predicting 

physiochemical properties of drugs has been validated by 

frameworks such as Chemi-Net and Molecule-Net, which have 

superior potential when compared to ML models [29].  

Role of Al in drug designing 

The most important aspect of drug designing is to predict the 

structure of the target protein in order to design the drug 

molecule. AI tools can assist by predicting the 3 dimensional 

(3D) protein structure, thus helping to predict the effect of a 

drug molecule on the target prior to its synthesis or production 

[30]. AlphaFold is an AI tool based on DNNs, used to scrutinize 

the distance between the adjacent amino acids and the 

correlating angles of the peptide bonds to predict the 3D target 

protein structure [17]. The development of AlphaFold is 

expected to revolutionize personalized medicine and drug 

discovery. It represents a significant step forward in the use of 

AI in structural biology. AI platforms such as Self-Organizing 

Map (SOM) are leveraged to link several compounds to 

numerous targets along with Bayesian classifiers and similarity 

ensemble approach (SEA) algorithms, which can be used for 

integration between the pharmacological profiles of drug 

molecules and their possible targets [31]. Newer methods based 

on NLP are implemented, which use the amino acid sequences 

from sequence databases to learn and predict structural and 

functional patterns. Evolutionary Scale Modeling-based 

software ESMfold was developed by Lin et al. in 2022, which 

utilizes a masked transformer protein language model with a 

deep understanding of biological properties, trained with over 

15 billion parameters [32].  

Apart from ML algorithms, DL models, such as Platform 

for Analytics and Distributed Machine Learning for Enterprises 

(PADME) and DeepAffinity, have the ability to predict drug-

target interactions by integrating drug and target features. 

PADME helps forecast the strength between drugs and target 

proteins, leading to accurate predictions of therapeutic efficacy 

and mechanism of action. DeepAffinity combines RNN and 

CNN with labeled and unlabeled data, providing a subtle 

understanding of drug-protein interactions [33, 34]. 

Apart from target protein prediction, the prediction of drug–

target binding affinity (DTBA) is critical for evaluating the 

efficacy of drug molecules. AI tools such as ChemMapper and 

SEA, along with ML and DL techniques such as SimBoost, 

DeepDTA, and PADME, have been employed to accurately 

predict DTBA [34]. These AI approaches have provided an 

upgrade to the traditional methods by deploying computational 

models for predicting interactions between drugs and protein 

targets. In recent times, DL methods have been used in de novo 

drug designing [35]. In 2018, Popova et al. developed a 

Reinforcement Learning (RL) algorithm for de novo drug 

synthesis, which involves employing generative and predictive 

DNNs to develop new drug molecules. The generative models 

are used to produce more unique molecules, whereas the 

predictive models foretell the properties of the developed 

compound [36]. The involvement of AI in the de novo drug 

design is beneficial to the pharmaceutical sector as it provides 

optimization of the already-learned data while suggesting 

possible synthesis routes for drug molecules, leading to prompt 

lead design and development [37, 38].  

Role of Al in clinical trials 

Clinical trials are necessary to prove the safety and efficacy of 

a drug for a particular disease condition and require anywhere 

from 6–7 years with substantial financial investments [17]. 

Clinical trial failure rates, especially in oncology and other 

diseases, can reach as high as 95% which further contributes to 

financial strain. Applying AI in various steps of clinical trials 

helps to improve patient stratification, intensify recruitment 

efficiency, and in the long run increase the likelihood of trial 

success [18]. In silico clinical trials (ISCT) are anticipated to 



Agrawal and Kulkarni                                                                          AI Applications in Drug Discovery and Development 

Online First                                                                                                                         Indian J Pharm Drug Studies | 4  

significantly reduce the costs of clinical trials while enhancing 

overall success rates [39]. ISCT integrates physiological and 

pathological data to generate patient-specific predictions that 

provide decisions regarding diagnosis, dose selection, 

prognosis, and the identification of suitable patient groups [40]. 

A newer concept, Virtual Physiological Human (VPH), 

configures virtual patient groups to test the safety and efficacy 

of new drugs and medical devices. VPH acts as an adjuvant to 

the traditional clinical trials by reducing the number of patients 

needed for the trials and enhancing the statistical power of the 

results [41]. Additionally, AI contributes to clinical trial 

success by linking patient genetic data and electronic health 

records, along with cumulating clinical trial databases to predict 

drug toxicity, and assisting with patient trial matching and 

recruitment, plus monitoring patient adherence during trials 

[18, 42]. 

One of the major causes of failure of clinical trials is 

toxicity. In light of this, PrOCTOR, a toxicity prediction model, 

was designed by Gayvert, et al. that helps to distinguish 

between FDA-approved drugs and drugs that failed during 

clinical trials due to toxicity. This enables the design of 

therapeutic agents with less toxicity [43]. TargeTox is another 

toxicity prediction model that produces and unites 

pharmacological and functional properties in an ML classifier 

to predict drug toxicity [44]. In order to increase patient 

adherence in clinical trials, AI-powered facial recognition 

algorithms are implemented, in which patients need to record a 

video of themselves swallowing the pills.  

The AI system then confirms that the correct person has 

taken the prescribed medication [45]. In a trial of patients with 

schizophrenia, adherence increased from 50% to 90% within a 

span of six months, exhibiting the success of AI-driven 

monitoring approaches [46]. A systematic review in 2024 

deduced that AI-derived drugs can have a success rate of 80–

90% in Phase I trials, around 40% in Phase II trials, which have 

shown to have substantially higher success rates than 

traditionally derived drugs [47]. In the forthcoming years, as 

more clinical results for AI-discovered drug molecules become 

available, it will be exhilarating to see how AI technologies will 

impact the pharmaceutical industry at the drug trial level. 

Table 2: AI tools/techniques employed in drug discovery [1, 

17] 

S. no Tool/Technique name Description  

1 Machine learning Predicts drug-target 
interactions, helps analyse 
biological activity, and 
optimizes lead molecular 
compounds 

2 Deep learning Helps in de novo drug 
designing, virtual screening, 
and predicting drug 
properties 

3 Reinforcement learning Applied to optimize drug 
combinations and dosages by 

taking multiple variables into 

consideration and 

maximizing desired 
outcomes 

4 Neural graph 
fingerprints 

Helps in various aspects of 
drug discovery, such as lead 
optimization, virtual 
screening, and drug 
properties’ prediction 

5 DeepChem Uses a Python-based AI 
system to find a suitable 
candidate in drug discovery 

6 AlphaFold Predicts 3D structures of 
target proteins 

7 PotentialNet Uses neural networks to 
predict binding affinity of 
ligands 

8 DeepTox Software that predicts the 
toxicity of around 12,000 
drugs 

Role of AI in drug repurposing  

The process of bringing new drugs to the market has significant 

hurdles in terms of time, labour, and cost. But finding new 

indications for an already existing drug can considerably reduce 

these costs by repurposing or repositioning it for other diseases 

[48]. Drug repurposing allows the drug in question to enter 

phase II and III clinical trials instantly, with notably lower 

development costs, as pharmacokinetic, pharmacodynamic, 

and toxicity profiles of the drug are already established [48].  

Reker, et al have developed a method called self-organizing 

map-based prediction of drug equivalence relationships (SPi-

DER). The model predicts molecular targets of known drugs, 

including key-target and off-target proteins [49]. Benevolent 

AI, an AI-enabled drug discovery company, utilizes AI tools to 

unwrap novel connections within vast, unstructured datasets 

with respect to drugs, and their clinical trial information, 

enabling drug repurposing and facilitating the discovery of 

valuable new indications of the same. Benevolent AI, in 

collaboration with Johnson & Johnson, has been redeveloping 

histamine H3 receptor inverse agonist, Bavisant, which was 

originally intended for attention deficit hyperactivity disorder 

(ADHD), and has been repositioned for the treatment of 

extreme daytime sleepiness in Parkinson’s disease [50, 51]. The 

synergy between AI and drug repurposing is therefore of 

paramount importance for addressing the unmet medical needs. 

Challenges and obstacles to using AI  

Despite the significant potential of AI in transforming the 

landscape of the pharmaceutical industry, several challenges 

are prevalent and need to be addressed. A few of the challenges 

include resource sustainability, the quality and suitability of 

data used to train models, and potential for bias, all of which 

could result in unequal access to medical treatment of certain 

groups of people, undermining the principles of equality and 

justice [25]. An example explaining the likelihood of bias is 

during drug target prediction, when the training dataset contains 

an excess amount of target data related to a ‘X’ disease while 

having insufficient data for the disease ‘Y’, the model may 



Agrawal and Kulkarni                                                                          AI Applications in Drug Discovery and Development 

Online First                                                                                                                         Indian J Pharm Drug Studies | 5  

predict targets for the ‘X’ more accurately, while predictions 

for ‘Y’ may be biased, thereby affecting the accuracy, 

specificity, and efficacy of a new drug development [18].  

The use of AI also raises concerns about job losses due to 

the emergence of automated technologies. Additionally, the use 

of AI in the pharmaceutical industry sparks debate about data 

privacy and security breaches. As AI systems depend on large 

amounts of data in order to function, the probability of sensitive 

personal information being accessed or misused increases. It is 

imperative that the collection and use of personal data be done 

in a way that respects the privacy of individuals and is in 

accordance with the relevant regulations [52]. 

Future perspectives  

As AI technology continues to evolve, its role in drug discovery 

and other aspects is expected to expand as well. A major 

breakthrough was sought recently when the pharmaceutical 

industry’s first AI-driven drug discovery, developed by Insilico 

Medicine, was approved by the FDA for clinical trials, which 

is in Phase II trials at the moment. The drug in question is 

Rentosertib, also known as ISM001-055, developed for the 

treatment of idiopathic pulmonary fibrosis (IPF) using their in-

house generative AI platform called Pharma.AI [52]. This 

development substantially paves the way for further expansion. 

Latest tools, such as quantum computing, could supplement 

AI’s computational abilities, enabling rapid and more precise 

predictions [20].  

It is very much possible that the future of AI-assisted drug 

discovery would pivot on developing a virtual human, allowing 

for accurate predictions of all interactions between drug 

molecules and exploring all therapeutic capabilities and adverse 

side effects [2]. Association between AI organizations, 

pharmaceutical companies, and policymakers will be vital for 

creating an ecosystem in which AI-powered drug discovery and 

development becomes the customary standard [20]. 

CONCLUSION 

Artificial Intelligence has a crucial role in the various stages of 

drug development, ranging from drug discovery, design, 

formulation, clinical trials, repositioning, and up to the final 

market introduction. Various tools assist in all the stages, 

enhancing efficiency and outcomes, while needing less time 

and man-power. Forging ahead, there is a need to strengthen 

data management, elucidate superior AI models, ensuring 

vigorous ethical and legal considerations, with the aim of 

sustainable and robust development of AI in the pharmaceutical 

industry. 

REFERENCES 

1. Ali, K.A., Mohin, S., Mondal, P. et al. Influence of artificial 
intelligence in modern pharmaceutical formulation and drug 
development. Futur J Pharm Sci. 2024;10:53.  
doi:10.1186/s43094-024-00625-1 

2. Blanco-González A, Cabezón A, Seco-González A, et al. The Role 
of AI in Drug Discovery: Challenges, Opportunities, and 

Strategies. Pharmaceuticals. 2023 Jun 18;16(6):891. doi: 

10.3390/ph16060891. PMID: 37375838; PMCID: 
PMC10302890. 

3. Khan SR, Baghdasarian A, Fahlman RP, et al. Current status and 

future prospects of toxicogenomics in drug discovery. Drug 
Discov. Today. 2014 May 1;19(5):562-
78.doi:10.1016/J.DRUDIS. 2013.11.001. 

4. J. A. DiMasi, H. G. Grabowski, R. W. Hansen. Integrated 
Continuous Pharmaceutical Technologies—A Review. J Health 
Econ. 2016;47:20-33. doi:10.1016/J.JHEALECO.2016.01.012. 

5. M. Bordukova, N. Makarov, R. Rodriguez-Esteban, et al. 
Generative artificial intelligence empowers digital twins in drug 

discovery and clinical trials. Expert Opin Drug Discov. 
2024;19:33–42. doi:10. 1080/17460441.2023.2273839. 

6. B. W. Wirtz, J. C. Weyerer, C. Geyer. Role of AI in drug 
discovery. IJPA. 2019;42:596–615. doi:10.1080/01900692. 
2018.1498103. 

7. Xu Y, Liu X, Cao X, et al. Artificial intelligence: A powerful 
paradigm for scientific research. Innovation. 2021 Oct 
28;2(4):100179. doi: 10.1016/j.xinn.2021.100179.  

8. Pu L, Naderi M, Liu T, Wu HC, et al. A machine learning-based 
approach to estimate the toxicity of drug candidates. BMC 
Pharmacol Toxicol. 2019 Jan 8;20(1):2. doi: 10.1186/s40360-018-
0282-6.  

9. M. J. Lamberti, M. Wilkinson, B. A. Donzanti, G. E., et al. A study 
on the application and use of artificial intelligence to support. drug 
development. Clin Ther. 2019;41:1414–1426. doi:10. 
1016/J.CLINTHERA.2019.05.018. 

10. S. Padte, V. Samala Venkata, P. Mehta, et al. 21st century critical 
care medicine: An overview, World J. Crit. Care Med. 2024. 
90176. doi: 10.5492/wjccm.v13.i1.90176. 

11. Sutiene K, Schwendner P, Sipos C, et al. Enhancing portfolio 
management using artificial intelligence: literature review. Front 
Artif Intell. 2024 Apr 8;7:1371502. doi: 
10.3389/frai.2024.1371502.  

12. Ding H, Tian J, Yu W, Wilson DI, et al. The Application of 
Artificial Intelligence and Big Data in the Food Industry. Foods. 

2023 Dec 18;12(24):4511. doi: 10.3390/foods12244511.  
13. Sierpe A, Yen RW, Stevens G, et al. Agenda-setting in the clinical 

encounter: A systematic review protocol. PLoS One. 2024 Oct 
24;19(10):e0312613. doi: 10.1371/journal.pone.0312613. 

14. González García C, Núñez Valdéz ER, García Díaz V, et al. A 
review of artificial intelligence in the internet of things. IJIMAI. 
2019;5:9–20. doi: 10.9781/ijimai.2018.03.004 

15. Bellini V, Rafano Carnà E, Russo M, et al. Artificial intelligence 

and anesthesia: a narrative review. Ann Transl Med. 
2022;10(9):528. doi: 10.21037/atm-21-7031.  

16. Bellini V, Cascella M, Cutugno F, et al. Understanding basic 
principles of Artificial Intelligence: a practical guide for 
intensivists. Acta Biomed. 2022 Oct 26;93(5):e2022297. doi: 
10.23750/abm.v93i5.13626.  

17. Paul D, Sanap G, Shenoy S, et al. Artificial intelligence in drug 
discovery and development. Drug Discov Today. 2021 

Jan;26(1):80-93. doi: 10.1016/j.drudis.2020.10.010.  
18. Fu C, Chen Q. The future of pharmaceuticals: Artificial 

intelligence in drug discovery and development. J. Pharm. Anal. 
2025 Feb 26:101248. doi: 10.1016/j.jpha.2025.101248 

19. Giammarile F, Paez D, Zimmermann R, et al. Production and 
regulatory issues for theranostics. Lancet Oncol. 2024 
Jun;25(6):e260-e269. doi: 10.1016/S1470-2045(24)00041-X.  

20. Nicole A. Colwell. Harnessing Artificial Intelligence in Drug 

Discovery and Development. ACCC. December 20 2024. 
21. Amorim AMB, Piochi LF, Gaspar AT, et al. Advancing Drug 

Safety in Drug Development: Bridging Computational Predictions 
for Enhanced Toxicity Prediction. Chem Res Toxicol. 2024 Jun 
17;37(6):827-849. doi: 10.1021/acs.chemrestox.3c00352 

22. Mak KK, Pichika MR. Artificial intelligence in drug development: 
present status and future prospects. Drug Discov. Today. 2019 Mar 
1;24(3):773-80. doi:10.1016/j.drudis.2018.11.014 

23. Zhang L, Tan J, Han D, et al. From machine learning to deep 
learning: progress in machine intelligence for rational drug 

https://doi.org/10.1186/s43094-024-00625-1
https://doi.org/10.1016/J.DRUDIS.%202013.11.001
https://doi.org/10.1016/j.jpha.2025.101248
https://doi.org/10.1016/j.drudis.2018.11.014


Agrawal and Kulkarni                                                                          AI Applications in Drug Discovery and Development 

Online First                                                                                                                         Indian J Pharm Drug Studies | 6  

discovery. Drug Discov Today. 2017 Nov;22(11):1680-1685. doi: 
10.1016/j.drudis.2017.08.010.  

24. Gentile F, Yaacoub JC, Gleave J, et al. Artificial intelligence-

enabled virtual screening of ultra-large chemical libraries with 
deep docking. Nat Protoc. 2022 Mar;17(3):672-697. doi: 
10.1038/s41596-021-00659-2.  

25. Rehman AU, Li M, Wu B, et al. Role of artificial intelligence in 
revolutionizing drug discovery. Fundam Res. 2024 May 
9;5(3):1273-1287. doi: 10.1016/j.fmre.2024.04.021.  

26. Carpenter KA, Huang X. Machine Learning-based Virtual 
Screening and Its Applications to Alzheimer's Drug Discovery: A 

Review. Curr Pharm Des. 2018;24(28):3347-3358. doi: 
10.2174/1381612824666180607124038.  

27. Zhu H. Big Data and Artificial Intelligence Modeling for Drug 
Discovery. Annu Rev Pharmacol Toxicol. 2020 Jan 6;60:573-589. 
doi: 10.1146/annurev-pharmtox-010919-023324.  

28. Cai H, Zhang H, Zhao D, et al. FP-GNN: a versatile deep learning 
architecture for enhanced molecular property prediction. Brief 
Bioinform. 2022 Nov 19;23(6):bbac408. doi: 

10.1093/bib/bbac408. 
29. Feinberg EN, Joshi E, Pande VS, et al. Improvement in ADMET 

Prediction with Multitask Deep Featurization. J Med Chem. 2020 
Aug 27;63(16):8835-8848. doi: 10.1021/acs.jmedchem.9b02187. 

30. Wan F, Zeng J. Deep learning with feature embedding for 
compound-protein interaction prediction. Biorxiv. 2016 Nov 
7:086033. doi: 10.1101/086033 

31. Achenbach J, Tiikkainen P, Franke L, et al. Computational tools 

for polypharmacology and repurposing. Future Med Chem. 2011 
Jun;3(8):961-8. doi: 10.4155/fmc.11.62.  

32. Tunyasuvunakool K, Adler J, Wu Z, et al. Highly accurate protein 
structure prediction for the human proteome. Nature. 2021 
Aug;596(7873):590-596. doi: 10.1038/s41586-021-03828-1.  

33. Feng Q, Dueva E, Cherkasov A, Ester M. Padme: A deep learning-
based framework for drug-target interaction prediction. arXiv 
preprint arXiv:1807.09741. 2018 Jul 
25.doi: 10.48550/arXiv.1807.09741 

34. Mostafa Karimi, Di Wu, Zhangyang Wang, et al. DeepAffinity: 
interpretable deep learning of compound–protein affinity through 
unified recurrent and convolutional neural networks. 
Bioinformatics. 2019;35(18):3329–3338. 
doi:10.1093/bioinformatics/btz111 

35. Hessler G, Baringhaus KH. Artificial Intelligence in Drug Design. 
Molecules. 2018 Oct 2;23(10):2520. doi: 
10.3390/molecules23102520.  

36. Popova M, Isayev O, Tropsha A. Deep reinforcement learning for 
de novo drug design. Sci Adv. 2018 Jul 25;4(7):eaap7885. doi: 
10.1126/sciadv.aap7885.  

37. Segler MHS, Kogej T, Tyrchan C, et al. Generating Focused 
Molecule Libraries for Drug Discovery with Recurrent Neural 
Networks. ACS Cent Sci. 2018 Jan 24;4(1):120-131. doi: 
10.1021/acscentsci.7b00512.  

38. Schneider G, Clark DE. Automated De Novo Drug Design: Are 

We Nearly There Yet? Angew Chem Int Ed Engl. 2019 Aug 
5;58(32):10792-10803. doi: 10.1002/anie.201814681. 

39. Creemers JHA, Ankan A, Roes KCB, et al. In silico cancer 
immunotherapy trials uncover the consequences of therapy-
specific response patterns for clinical trial design and outcome. Nat 
Commun. 2023 Apr 24;14(1):2348. doi: 10.1038/s41467-023-
37933-8.  

40. Ramella A, Migliavacca F, Rodriguez Matas JF, et al. 

Applicability assessment for in-silico patient-specific TEVAR 
procedures. J Biomech. 2023 Jan;146:111423. doi: 
10.1016/j.jbiomech.2022.111423.  

41. Musuamba FT, Skottheim Rusten I, Lesage R, et al. Scientific and 
regulatory evaluation of mechanistic in silico drug and disease 
models in drug development: Building model credibility. CPT 

Pharmacometrics Syst Pharmacol. 2021 Aug;10(8):804-825. doi: 
10.1002/psp4.12669.  

42. Giaretta A, Petrucci G, Rocca B, et al. Physiologically based 
modelling of the antiplatelet effect of aspirin: A tool to 
characterize drug responsiveness and inform precision dosing. 
PLoS One. 2022 Aug 17;17(8):e0268905. doi: 
10.1371/journal.pone.0268905. 

43. Gayvert KM, Madhukar NS, Elemento O. A Data-Driven 

Approach to Predicting Successes and Failures of Clinical Trials. 
Cell Chem Biol. 2016 Oct 20;23(10):1294-1301. doi: 
10.1016/j.chembiol.2016.07.023. 

44. Lysenko A, Sharma A, Boroevich KA, et al. An integrative 
machine learning approach for prediction of toxicity-related drug 
safety. Life Sci Alliance. 2018 Nov 28;1(6):e201800098. doi: 
10.26508/lsa.201800098.  

45. Labovitz DL, Shafner L, Reyes Gil M, et al. Using Artificial 

Intelligence to Reduce the Risk of Nonadherence in Patients on 
Anticoagulation Therapy. Stroke. 2017 May;48(5):1416-1419. 
doi: 10.1161/STROKEAHA.116.016281. 

46. Bain EE, Shafner L, Walling DP, et al. Use of a Novel Artificial 
Intelligence Platform on Mobile Devices to Assess Dosing 
Compliance in a Phase 2 Clinical Trial in Subjects With 
Schizophrenia. JMIR Mhealth Uhealth. 2017 Feb 21;5(2):e18. doi: 
10.2196/mhealth.7030.  

47. Kp Jayatunga M, Ayers M, Bruens L, et al. How successful are AI-
discovered drugs in clinical trials? A first analysis and emerging 
lessons. Drug Discov Today. 2024 Jun;29(6):104009. doi: 
10.1016/j.drudis.2024.104009.  

48. Datti A. Academic drug discovery in an age of research 
abundance, and the curious case of chemical screens toward drug 
repositioning. Drug Discov Today. 2023 May;28(5):103522. doi: 
10.1016/j.drudis.2023.103522.  

49. Reker D, Rodrigues T, Schneider P, et al. Identifying the 

macromolecular targets of de novo-designed chemical entities 
through self-organizing map consensus. Proc Natl Acad Sci U S 
A. 2014 Mar 18;111(11):4067-72. doi: 10.1073/pnas.1320001111.  

50. Watanabe H, Hattori T, Kume A, et al. Improved Parkinsons 
disease motor score in a single-arm open-label trial of febuxostat 
and inosine. Medicine (Baltimore). 2020 Aug 28;99(35):e21576. 
doi: 10.1097/MD.0000000000021576.  

51. van der Pol KH, Aljofan M, Blin O, et al. Drug Repurposing of 

Generic Drugs: Challenges and the Potential Role for Government. 
Appl Health Econ Health Policy. 2023 Nov;21(6):831-840. doi: 
10.1007/s40258-023-00816-6.  

52. Xu Z, Ren F, Wang P, et al. A generative AI-discovered TNIK 
inhibitor for idiopathic pulmonary fibrosis: a randomized phase 2a 
trial. Nat Med. 2025 Jun 3:1-9. doi:10.1038/s41591-025-03743-2. 

 

 

How to cite this article: Agrawal R, Kulkarni M. 
Application of Artificial Intelligence in drug discovery, 

designing, clinical trials and repurposing. Indian J Pharm 

Drug Studies. 2025; Online First. 

Funding: None;                 Conflicts of Interest: None Stated 

 

http://dx.doi.org/10.48550/arXiv.1807.09741
https://doi.org/10.1093/bioinformatics/btz111

