Deep Learning to Artificial Intelligence: A Machine Intelligence Approach to Drug Discovery
Abstract
The paper explores the pivotal role of artificial intelligence (AI) and machine learning (ML) in drug discovery and development. It delves into the evolution of AI from ML to deep learning and its application across various stages of the drug discovery process. The article underscores the challenges in traditional drug discovery and how AI can surmount them, providing an overview of AI applications in drug design and discovery, encompassing virtual screening, toxicity prediction, drug monitoring, and drug repositioning. It also delves into the history and classification of AI in healthcare and pharmaceutical industries. The comprehensive list of references further underscores the significance of AI and deep learning in drug discovery. The use of AI and ML in drug discovery is crucial for identifying potential drug targets, finding suitable drug-like molecules, and improving drug development processes.
AI has been integrated with traditional chemistry to enhance drug discovery and development, transforming computational drug design through techniques such as predicting protein structures and using quantum mechanics. AI has also been applied in various stages of drug development, including peptide synthesis, small molecule design, drug dosage determination, and drug delivery effectiveness. Additionally, AI algorithms have been used to optimize drug dosage, identify potential lead compounds, predict adverse drug reactions, and conduct virtual screening for drug discovery, thereby speeding up the drug discovery process and making it more efficient and reliable. (Vodanović et al.,2021)
The application of AI in structure and ligand-based virtual screening, QSAR modeling, and drug repositioning has also been discussed, demonstrating the potential of AI-based tools in drug discovery. The paper also provides a comprehensive list of tools and software used in drug discovery and development, highlighting the significant role of AI in predicting physicochemical properties. AI algorithms have been implemented in drug design and discovery processes, including QSAR modeling, drug repositioning, and predicting physicochemical properties and bioactivity of compounds. The paper emphasizes the potential of AI to reduce drug development time, cost, attrition rates, and human resources, making it a valuable tool in the pharmaceutical industry. (Gupta et al.,2021)
The use of AI in drug discovery and development, particularly in the context of neurodegenerative diseases, has been discussed, covering various AI techniques such as machine learning, deep learning, and neural networks, and their applications in target identification, hit discovery, lead optimization, and clinical trial design. The paper also addresses the challenges and potential solutions in using AI in drug discovery, acknowledging the need for standard evaluation and addressing legal and ethical concerns.
Introduction
1.1 Background
The field of drug discovery faces significant challenges, including high costs, lengthy timelines, and high attrition rates in clinical trials. These challenges have prompted the exploration of innovative approaches to streamline the drug design and development process. Artificial intelligence (AI) and machine learning (ML) have emerged as promising technologies to address these challenges by enhancing the efficiency and effectiveness of drug discovery.(Vodanović et al.,2023)
1.2 Motivation
The motivation for innovative approaches in drug design and development stems from the urgent need to expedite the discovery of novel therapeutics, improve the success rate of clinical trials, and lower the total cost of introducing novel medications to the market. Traditional drug discovery methods have limitations in terms of speed and accuracy, making it imperative to explore advanced technologies such as AI and deep learning to revolutionize the process.
1.3 Objectives
The primary objective of this survey paper is to provide a comprehensive overview of the role of AI and deep learning in drug discovery and development. This includes examining their applications in various stages of the drug discovery process, such as virtual screening, toxicity prediction, drug monitoring, and drug repositioning. Additionally, the paper aims to highlight the potential of AI to address the challenges faced in traditional drug discovery and to provide insights into the future prospects of AI in the pharmaceutical industry.
Drug Discovery Challenges
2.1 Low Efficacy and Off-Target Delivery
One of the significant challenges in drug discovery is the low efficacy and off-target delivery of drugs. The development of effective medications against diseases is hindered by the difficulty in ensuring that drugs specifically target the intended site of action without causing adverse effects in off-target areas.
This challenge is further compounded by the complex nature of biological systems, where the action of drugs is intricate and often unpredictable. Additionally, the pharmaceutical industry faces the issue of limited data availability for training AI models, as well as the reluctance to share pharmacokinetic and pharmacodynamic measurements of drugs until they are approved. This lack of data hampers the ability to develop accurate and reliable AI models for drug discovery .
2.2 Time Consumption and Cost
The drug discovery process is characterized by extensive time consumption and high costs. The various stages involved, including pre-clinical and clinical trials, as well as manufacturing practices, contribute to the prolonged timelines and substantial financial investments required. Managing the cost and speed of the drug discovery process is a significant concern for pharmaceutical companies.
However, the implementation of AI has shown promise in reducing the time consumption and cost of the process through its ability to streamline and optimize various aspects of drug discovery, including virtual screening, toxicity prediction, and drug repositioning .
2.3 Big Data Challenges
The exponential growth of biological and chemical data from genomics studies, proteomics studies, microarray data, and clinical trials has led to the generation of vast amounts of big data. However, the effective utilization of this data presents a significant challenge in drug discovery. The curation and analysis of this data for pharmacological and physicochemical properties necessary for the drug discovery process require advanced computational techniques, including artificial intelligence and machine learning.
Furthermore, the filtration and analysis of the screened data, followed by the application of AI models such as deep learning, random forest, and neural networks, present challenges in terms of data processing and interpretation. Additionally, the integration of quantum mechanical properties in the drug discovery process poses further challenges, as these properties play a crucial role but cannot directly hamper the drug designing process .
In conclusion, the challenges in drug discovery related to low efficacy and off-target delivery, time consumption and cost, and big data pose significant hurdles for the pharmaceutical industry. However, the integration of AI and advanced computational techniques presents opportunities to address these challenges and revolutionize the drug discovery process.(Blanco-González et al.,2023)
Role of Artificial Intelligence and Machine Learning
3.1 Overview
In the field of drug development, artificial intelligence (AI) and machine learning (ML) have completely changed the game. Machine learning (ML) is a subset of artificial intelligence (AI) that allows systems to automatically learn from experience and get better without explicit programming. AI is the simulation of human intelligence in computers. In drug discovery, AI and ML have evolved from traditional computational methods to advanced deep learning techniques, offering innovative solutions to the challenges faced in the pharmaceutical industry.
AI and ML have significantly enhanced the drug discovery process by streamlining various stages, including target identification, lead optimization, and clinical trial design. These technologies have the potential to expedite the discovery of novel therapeutics, improve the success rate of clinical trials, and reduce the overall cost of bringing new drugs to market. The integration of AI and ML in drug discovery has led to the development of predictive models for drug-target interactions, toxicity prediction, and virtual screening, thereby accelerating the identification of potential drug candidates.(Qureshi et al.,2023)
3.2 Application Areas
The application of machine learning and deep learning in drug discovery spans various areas, including virtual screening, toxicity prediction, drug monitoring, and drug repositioning. Virtual screening, a critical stage in drug discovery, involves the use of computational methods to identify potential drug candidates from large compound libraries. Machine learning algorithms have been instrumental in enhancing the accuracy and efficiency of virtual screening by predicting the binding affinity of drug molecules with therapeutic targets and evaluating drug toxicity.
Toxicity prediction is another crucial area where AI and ML have been applied in drug discovery. These technologies enable the early identification of potential adverse effects of drug candidates, thereby reducing the risk of failure in later stages of development. Additionally, AI and ML play a significant role in drug monitoring, aiding in the optimization of drug dosage and the identification of bioactive compounds.
Furthermore, AI and ML have been utilized in drug repositioning, which involves identifying new therapeutic uses for existing drugs. By analyzing large-scale biological and chemical data, these technologies can uncover potential drug candidates for the treatment of different diseases, thereby expediting the drug development process.
Implementation in Drug Discovery Processes
4.1 Peptide Synthesis
The application of artificial intelligence (AI) in peptide synthesis has significantly advanced the drug discovery process. AI algorithms have been utilized to predict and optimize peptide sequences, design novel peptide structures, and enhance the efficiency of peptide synthesis. By leveraging machine learning (ML) and deep learning techniques, AI has enabled the rapid and accurate prediction of peptide properties, such as bioactivity, stability, and solubility. Additionally, AI-driven peptide synthesis platforms have facilitated the identification of potential therapeutic peptides and the development of peptide-based drugs. These platforms utilize predictive models to optimize the synthesis of peptides with desired properties, thereby accelerating the discovery of novel peptide-based therapeutics.
4.2 Structure-Based Virtual Screening
Artificial intelligence plays a crucial role in structure-based virtual screening (SBVS) for drug discovery. AI algorithms, including machine learning and deep learning, have been instrumental in accurately predicting the binding affinity of small molecules with target proteins. By leveraging structural information of target proteins and small molecules, AI-based SBVS methods enable the efficient identification of potential drug candidates. These algorithms have significantly improved the accuracy and speed of virtual screening, leading to the discovery of novel lead compounds with high binding affinity and specificity for therapeutic targets.
4.3 Ligand-Based Virtual Screening
The application of AI in ligand-based virtual screening (LBVS) has revolutionized the drug discovery process. AI algorithms have been employed to analyze large chemical databases and predict the bioactivity and pharmacological properties of small molecules. By utilizing machine learning and deep learning techniques, AI-based LBVS methods can efficiently identify potential drug candidates based on their structural and physicochemical properties. These algorithms have enhanced the accuracy and reliability of virtual screening, enabling the rapid identification of bioactive compounds for further development.
4.4 Other Processes (Toxicity Prediction, Drug Monitoring, etc.)
AI and deep learning have impacted various other drug discovery processes, including toxicity prediction, drug monitoring, and drug repositioning. AI algorithms have been utilized to predict the toxicity of drug candidates, enabling the early identification of potential adverse effects and reducing the risk of late-stage failures. Additionally, AI-based models have facilitated drug monitoring by optimizing drug dosage and identifying bioactive compounds. Furthermore, AI-driven drug repositioning approaches have leveraged large-scale biological and chemical data to identify new therapeutic uses for existing drugs, accelerating the discovery of novel indications and expanding the potential applications of existing medications.(You et al.,2022)
Table : The application of AI in different Drug discovery.
Evidence and Advancements:
5.1 Past Implementations
Previous studies have provided compelling evidence supporting the use of AI in drug discovery. For instance, AI and multi-objective optimization have been shown to capture the latent links between chemical and biological aspects, providing customizable design strategies for lead generation and lead optimization .
Machine learning models, such as support vector machines (SVM), random forests (RF), and deep neural networks (DNNs), have been successfully applied in drug discovery for analyzing pharmaceutical applications from docking to virtual screening . Additionally, AI-driven drug repurposing approaches have demonstrated effectiveness in minimizing drug development duration through data mining and AI techniques . These past implementations have showcased the potential of AI in revolutionizing the drug discovery process, from lead identification to clinical development .
Furthermore, a recent success story in AI-driven drug discovery involves the proposal of a novel target and its inhibitor through AI-based tools for the treatment of idiopathic pulmonary fibrosis. In silico medicine, a biotechnology company, utilized AI-based tools to identify a small molecule inhibitor with promising efficacy in human cells and animal models, leading to its nomination for investigational new drug (IND) enabling studies and subsequent clinical trials .(Gupta et al,2021)
5.2 Recent Developments
Recent advancements in data mining, curation, and management techniques have provided critical support to modeling algorithms in drug discovery. Novel data mining approaches, such as question-answer artificial systems (QAAI) utilizing Google semantic AI universal encoder, have been effective in drug repurposing, as demonstrated by the successful prediction of the lipoxygenase inhibitor drug zileuton as a modulator of the NRF2 pathway . These developments have facilitated the identification of new therapeutic uses for existing drugs, thereby accelerating the drug discovery process.
Moreover, advancements in data curation and management have enhanced the integration of AI and deep learning in drug discovery. Novel data mining techniques have provided critical support to recently developed modeling algorithms, enabling the efficient implementation of AI and deep learning in various drug discovery processes, including peptide synthesis, virtual screening, toxicity prediction, and drug repositioning .
In summary, past implementations and recent developments in AI-driven drug discovery have demonstrated the potential of AI and deep learning in revolutionizing the drug discovery process, from lead identification to clinical development. These advancements have paved the way for the integration of AI and deep learning in various drug discovery processes, offering promising prospects for the development of novel therapeutics.(Vodanović et al. , 2023)
Opportunities and Impact:
6.1 Rational Drug Design
AI and deep learning offer significant opportunities for rational drug design. These technologies enable the efficient analysis of large datasets, including genomic, proteomic, and chemical information, to identify potential drug targets and design novel therapeutics . By leveraging machine learning algorithms, AI can predict the binding affinity of small molecules with target proteins, facilitating the rational design of drug candidates with enhanced specificity and efficacy . Furthermore, deep learning techniques have been instrumental in the de novo design of drug-like molecules, enabling the generation of novel chemical structures with desired pharmacological properties . These advancements in AI-driven rational drug design have the potential to expedite the discovery of innovative therapeutics and optimize the drug development process .
6.2 Impact on Mankind
The potential impact of AI advancements on drug discovery is profound and far-reaching. AI-driven drug discovery has the potential to revolutionize the pharmaceutical industry by accelerating the development of novel therapeutics, improving the success rate of clinical trials, and reducing the overall cost of bringing new drugs to market . By leveraging AI and deep learning, researchers can efficiently identify potential drug candidates, predict their pharmacological properties, and optimize their therapeutic efficacy, ultimately leading to the development of more effective and personalized treatments for various diseases . Furthermore, AI-driven drug discovery has the potential to address unmet medical needs, expedite the identification of new therapeutic uses for existing drugs, and enhance the efficiency of the drug development process . The implications of AI advancements in drug discovery for humanity are significant, as these technologies have the potential to improve global healthcare outcomes, extend life expectancy, and mitigate the burden of disease on society.
Conclusion
The survey provides a comprehensive overview of the application of artificial intelligence (AI) and machine learning (ML) in drug discovery and development. It highlights the evolution of AI from ML to deep learning and its application in various stages of the drug discovery process. The survey covers various stages of the drug development process, including peptide synthesis, small molecule design, drug dosage determination, and drug delivery effectiveness, showcasing the role of AI in streamlining and improving the drug discovery process.
Furthermore, the survey discusses the use of AI in structure and ligand-based virtual screening, quantitative structure-activity relationship (QSAR) modeling, drug repositioning, and predicting physicochemical properties and bioactivity of compounds. It also emphasizes the development of AI-based tools and software packages to assist in predicting drug-target interactions, evaluating drug toxicity, and identifying molecular pathways and polypharmacology. The survey acknowledges the potential of AI to revolutionize drug discovery and development, particularly in the context of neurodegenerative diseases, and highlights the challenges and opportunities associated with the use of AI in the pharmaceutical industry.The survey also provides a comprehensive list of tools and software used in drug discovery and development, including QSAR modeling, drug repurposing, prediction of drug-target interactions, prediction of toxicity, and prediction of physicochemical properties. These tools utilize machine learning, deep learning, neural networks, and other computational methods to integrate diverse chemical and biological data, predict drug-disease associations, and assess the mode of action and toxicity of compounds. The use of AI-based approaches in drug discovery and development is highlighted as a significant factor in predicting physicochemical properties.Overall, the survey contributes to the understanding of the significant role of AI and ML in advancing drug discovery and development, showcasing their potential to accelerate the process, improve efficiency, and revolutionize pharmaceutical research and healthcare. The survey also references various studies and developments in the field of AI and drug discovery, providing a comprehensive list of research papers and articles related to drug discovery, computational biology, and artificial intelligence in healthcare. These references cover a wide range of topics and applications within the field of pharmaceutical research and development, further emphasizing the impact and potential of AI in drug discovery and development.
References
Abstract
The paper explores the pivotal role of artificial intelligence (AI) and machine learning (ML) in drug discovery and development. It delves into the evolution of AI from ML to deep learning and its application across various stages of the drug discovery process. The article underscores the challenges in traditional drug discovery and how AI can surmount them, providing an overview of AI applications in drug design and discovery, encompassing virtual screening, toxicity prediction, drug monitoring, and drug repositioning. It also delves into the history and classification of AI in healthcare and pharmaceutical industries. The comprehensive list of references further underscores the significance of AI and deep learning in drug discovery. The use of AI and ML in drug discovery is crucial for identifying potential drug targets, finding suitable drug-like molecules, and improving drug development processes.
AI has been integrated with traditional chemistry to enhance drug discovery and development, transforming computational drug design through techniques such as predicting protein structures and using quantum mechanics. AI has also been applied in various stages of drug development, including peptide synthesis, small molecule design, drug dosage determination, and drug delivery effectiveness. Additionally, AI algorithms have been used to optimize drug dosage, identify potential lead compounds, predict adverse drug reactions, and conduct virtual screening for drug discovery, thereby speeding up the drug discovery process and making it more efficient and reliable. (Vodanović et al.,2021)
The application of AI in structure and ligand-based virtual screening, QSAR modeling, and drug repositioning has also been discussed, demonstrating the potential of AI-based tools in drug discovery. The paper also provides a comprehensive list of tools and software used in drug discovery and development, highlighting the significant role of AI in predicting physicochemical properties. AI algorithms have been implemented in drug design and discovery processes, including QSAR modeling, drug repositioning, and predicting physicochemical properties and bioactivity of compounds. The paper emphasizes the potential of AI to reduce drug development time, cost, attrition rates, and human resources, making it a valuable tool in the pharmaceutical industry. (Gupta et al.,2021)
The use of AI in drug discovery and development, particularly in the context of neurodegenerative diseases, has been discussed, covering various AI techniques such as machine learning, deep learning, and neural networks, and their applications in target identification, hit discovery, lead optimization, and clinical trial design. The paper also addresses the challenges and potential solutions in using AI in drug discovery, acknowledging the need for standard evaluation and addressing legal and ethical concerns.
Introduction
1.1 Background
The field of drug discovery faces significant challenges, including high costs, lengthy timelines, and high attrition rates in clinical trials. These challenges have prompted the exploration of innovative approaches to streamline the drug design and development process. Artificial intelligence (AI) and machine learning (ML) have emerged as promising technologies to address these challenges by enhancing the efficiency and effectiveness of drug discovery.(Vodanović et al.,2023)
1.2 Motivation
The motivation for innovative approaches in drug design and development stems from the urgent need to expedite the discovery of novel therapeutics, improve the success rate of clinical trials, and lower the total cost of introducing novel medications to the market. Traditional drug discovery methods have limitations in terms of speed and accuracy, making it imperative to explore advanced technologies such as AI and deep learning to revolutionize the process.
1.3 Objectives
The primary objective of this survey paper is to provide a comprehensive overview of the role of AI and deep learning in drug discovery and development. This includes examining their applications in various stages of the drug discovery process, such as virtual screening, toxicity prediction, drug monitoring, and drug repositioning. Additionally, the paper aims to highlight the potential of AI to address the challenges faced in traditional drug discovery and to provide insights into the future prospects of AI in the pharmaceutical industry.
Drug Discovery Challenges
2.1 Low Efficacy and Off-Target Delivery
One of the significant challenges in drug discovery is the low efficacy and off-target delivery of drugs. The development of effective medications against diseases is hindered by the difficulty in ensuring that drugs specifically target the intended site of action without causing adverse effects in off-target areas.
This challenge is further compounded by the complex nature of biological systems, where the action of drugs is intricate and often unpredictable. Additionally, the pharmaceutical industry faces the issue of limited data availability for training AI models, as well as the reluctance to share pharmacokinetic and pharmacodynamic measurements of drugs until they are approved. This lack of data hampers the ability to develop accurate and reliable AI models for drug discovery .
2.2 Time Consumption and Cost
The drug discovery process is characterized by extensive time consumption and high costs. The various stages involved, including pre-clinical and clinical trials, as well as manufacturing practices, contribute to the prolonged timelines and substantial financial investments required. Managing the cost and speed of the drug discovery process is a significant concern for pharmaceutical companies.
However, the implementation of AI has shown promise in reducing the time consumption and cost of the process through its ability to streamline and optimize various aspects of drug discovery, including virtual screening, toxicity prediction, and drug repositioning .
2.3 Big Data Challenges
The exponential growth of biological and chemical data from genomics studies, proteomics studies, microarray data, and clinical trials has led to the generation of vast amounts of big data. However, the effective utilization of this data presents a significant challenge in drug discovery. The curation and analysis of this data for pharmacological and physicochemical properties necessary for the drug discovery process require advanced computational techniques, including artificial intelligence and machine learning.
Furthermore, the filtration and analysis of the screened data, followed by the application of AI models such as deep learning, random forest, and neural networks, present challenges in terms of data processing and interpretation. Additionally, the integration of quantum mechanical properties in the drug discovery process poses further challenges, as these properties play a crucial role but cannot directly hamper the drug designing process .
In conclusion, the challenges in drug discovery related to low efficacy and off-target delivery, time consumption and cost, and big data pose significant hurdles for the pharmaceutical industry. However, the integration of AI and advanced computational techniques presents opportunities to address these challenges and revolutionize the drug discovery process.(Blanco-González et al.,2023)
Role of Artificial Intelligence and Machine Learning
3.1 Overview
In the field of drug development, artificial intelligence (AI) and machine learning (ML) have completely changed the game. Machine learning (ML) is a subset of artificial intelligence (AI) that allows systems to automatically learn from experience and get better without explicit programming. AI is the simulation of human intelligence in computers. In drug discovery, AI and ML have evolved from traditional computational methods to advanced deep learning techniques, offering innovative solutions to the challenges faced in the pharmaceutical industry.
AI and ML have significantly enhanced the drug discovery process by streamlining various stages, including target identification, lead optimization, and clinical trial design. These technologies have the potential to expedite the discovery of novel therapeutics, improve the success rate of clinical trials, and reduce the overall cost of bringing new drugs to market. The integration of AI and ML in drug discovery has led to the development of predictive models for drug-target interactions, toxicity prediction, and virtual screening, thereby accelerating the identification of potential drug candidates.(Qureshi et al.,2023)
3.2 Application Areas
The application of machine learning and deep learning in drug discovery spans various areas, including virtual screening, toxicity prediction, drug monitoring, and drug repositioning. Virtual screening, a critical stage in drug discovery, involves the use of computational methods to identify potential drug candidates from large compound libraries. Machine learning algorithms have been instrumental in enhancing the accuracy and efficiency of virtual screening by predicting the binding affinity of drug molecules with therapeutic targets and evaluating drug toxicity.
Toxicity prediction is another crucial area where AI and ML have been applied in drug discovery. These technologies enable the early identification of potential adverse effects of drug candidates, thereby reducing the risk of failure in later stages of development. Additionally, AI and ML play a significant role in drug monitoring, aiding in the optimization of drug dosage and the identification of bioactive compounds.
Furthermore, AI and ML have been utilized in drug repositioning, which involves identifying new therapeutic uses for existing drugs. By analyzing large-scale biological and chemical data, these technologies can uncover potential drug candidates for the treatment of different diseases, thereby expediting the drug development process.
Implementation in Drug Discovery Processes
4.1 Peptide Synthesis
The application of artificial intelligence (AI) in peptide synthesis has significantly advanced the drug discovery process. AI algorithms have been utilized to predict and optimize peptide sequences, design novel peptide structures, and enhance the efficiency of peptide synthesis. By leveraging machine learning (ML) and deep learning techniques, AI has enabled the rapid and accurate prediction of peptide properties, such as bioactivity, stability, and solubility. Additionally, AI-driven peptide synthesis platforms have facilitated the identification of potential therapeutic peptides and the development of peptide-based drugs. These platforms utilize predictive models to optimize the synthesis of peptides with desired properties, thereby accelerating the discovery of novel peptide-based therapeutics.
4.2 Structure-Based Virtual Screening
Artificial intelligence plays a crucial role in structure-based virtual screening (SBVS) for drug discovery. AI algorithms, including machine learning and deep learning, have been instrumental in accurately predicting the binding affinity of small molecules with target proteins. By leveraging structural information of target proteins and small molecules, AI-based SBVS methods enable the efficient identification of potential drug candidates. These algorithms have significantly improved the accuracy and speed of virtual screening, leading to the discovery of novel lead compounds with high binding affinity and specificity for therapeutic targets.
4.3 Ligand-Based Virtual Screening
The application of AI in ligand-based virtual screening (LBVS) has revolutionized the drug discovery process. AI algorithms have been employed to analyze large chemical databases and predict the bioactivity and pharmacological properties of small molecules. By utilizing machine learning and deep learning techniques, AI-based LBVS methods can efficiently identify potential drug candidates based on their structural and physicochemical properties. These algorithms have enhanced the accuracy and reliability of virtual screening, enabling the rapid identification of bioactive compounds for further development.
4.4 Other Processes (Toxicity Prediction, Drug Monitoring, etc.)
AI and deep learning have impacted various other drug discovery processes, including toxicity prediction, drug monitoring, and drug repositioning. AI algorithms have been utilized to predict the toxicity of drug candidates, enabling the early identification of potential adverse effects and reducing the risk of late-stage failures. Additionally, AI-based models have facilitated drug monitoring by optimizing drug dosage and identifying bioactive compounds. Furthermore, AI-driven drug repositioning approaches have leveraged large-scale biological and chemical data to identify new therapeutic uses for existing drugs, accelerating the discovery of novel indications and expanding the potential applications of existing medications.(You et al.,2022)
Table : The application of AI in different Drug discovery.
| Drug Discovery Process | Key Techniques | Advantages | Notable Studies |
| Virtual Screening | Structure-based virtual screening (SBVS), Ligand-based virtual screening (LBVS), AI algorithms | Efficient identification of potential lead compounds, Reduced time and cost in screening, Prediction of binding affinity | "Deep Learning for Virtual Screening of Potential Inhibitors for COVID-19" |
| Toxicity Prediction | Machine learning, Deep learning | Early identification of potential adverse effects, Reduction in animal testing, Improved safety profile of drugs | "Prediction of Antiviral Drugs for SARS-CoV-2 Using Deep Learning Frameworks" |
| Drug Repositioning | Data mining, AI algorithms | Accelerated identification of new therapeutic uses for existing drugs, Reduced development time and cost | "AI-Driven Drug Repurposing for Idiopathic Pulmonary Fibrosis Treatment" |
| QSAR Modeling | Quantitative Structure-Activity Relationship (QSAR) modeling, Machine learning algorithms | Prediction of physicochemical properties and bioactivity of compounds, Optimization of drug design | "Application of QSAR Modeling in Predicting Drug-Target Interactions" |
| Target Identification | Machine learning, Neural networks | Efficient identification of potential drug targets, Accelerated lead optimization | "AI Applications in Target Identification for Neurodegenerative Diseases" |
| Clinical Trial Design | AI applications, Machine learning | Improved patient recruitment, Enhanced trial efficiency, Reduced attrition rates | "AI Applications in Clinical Trial Design for Neurodegenerative Diseases" |
5.1 Past Implementations
Previous studies have provided compelling evidence supporting the use of AI in drug discovery. For instance, AI and multi-objective optimization have been shown to capture the latent links between chemical and biological aspects, providing customizable design strategies for lead generation and lead optimization .
Machine learning models, such as support vector machines (SVM), random forests (RF), and deep neural networks (DNNs), have been successfully applied in drug discovery for analyzing pharmaceutical applications from docking to virtual screening . Additionally, AI-driven drug repurposing approaches have demonstrated effectiveness in minimizing drug development duration through data mining and AI techniques . These past implementations have showcased the potential of AI in revolutionizing the drug discovery process, from lead identification to clinical development .
Furthermore, a recent success story in AI-driven drug discovery involves the proposal of a novel target and its inhibitor through AI-based tools for the treatment of idiopathic pulmonary fibrosis. In silico medicine, a biotechnology company, utilized AI-based tools to identify a small molecule inhibitor with promising efficacy in human cells and animal models, leading to its nomination for investigational new drug (IND) enabling studies and subsequent clinical trials .(Gupta et al,2021)
5.2 Recent Developments
Recent advancements in data mining, curation, and management techniques have provided critical support to modeling algorithms in drug discovery. Novel data mining approaches, such as question-answer artificial systems (QAAI) utilizing Google semantic AI universal encoder, have been effective in drug repurposing, as demonstrated by the successful prediction of the lipoxygenase inhibitor drug zileuton as a modulator of the NRF2 pathway . These developments have facilitated the identification of new therapeutic uses for existing drugs, thereby accelerating the drug discovery process.
Moreover, advancements in data curation and management have enhanced the integration of AI and deep learning in drug discovery. Novel data mining techniques have provided critical support to recently developed modeling algorithms, enabling the efficient implementation of AI and deep learning in various drug discovery processes, including peptide synthesis, virtual screening, toxicity prediction, and drug repositioning .
In summary, past implementations and recent developments in AI-driven drug discovery have demonstrated the potential of AI and deep learning in revolutionizing the drug discovery process, from lead identification to clinical development. These advancements have paved the way for the integration of AI and deep learning in various drug discovery processes, offering promising prospects for the development of novel therapeutics.(Vodanović et al. , 2023)
Opportunities and Impact:
6.1 Rational Drug Design
AI and deep learning offer significant opportunities for rational drug design. These technologies enable the efficient analysis of large datasets, including genomic, proteomic, and chemical information, to identify potential drug targets and design novel therapeutics . By leveraging machine learning algorithms, AI can predict the binding affinity of small molecules with target proteins, facilitating the rational design of drug candidates with enhanced specificity and efficacy . Furthermore, deep learning techniques have been instrumental in the de novo design of drug-like molecules, enabling the generation of novel chemical structures with desired pharmacological properties . These advancements in AI-driven rational drug design have the potential to expedite the discovery of innovative therapeutics and optimize the drug development process .
6.2 Impact on Mankind
The potential impact of AI advancements on drug discovery is profound and far-reaching. AI-driven drug discovery has the potential to revolutionize the pharmaceutical industry by accelerating the development of novel therapeutics, improving the success rate of clinical trials, and reducing the overall cost of bringing new drugs to market . By leveraging AI and deep learning, researchers can efficiently identify potential drug candidates, predict their pharmacological properties, and optimize their therapeutic efficacy, ultimately leading to the development of more effective and personalized treatments for various diseases . Furthermore, AI-driven drug discovery has the potential to address unmet medical needs, expedite the identification of new therapeutic uses for existing drugs, and enhance the efficiency of the drug development process . The implications of AI advancements in drug discovery for humanity are significant, as these technologies have the potential to improve global healthcare outcomes, extend life expectancy, and mitigate the burden of disease on society.
Conclusion
The survey provides a comprehensive overview of the application of artificial intelligence (AI) and machine learning (ML) in drug discovery and development. It highlights the evolution of AI from ML to deep learning and its application in various stages of the drug discovery process. The survey covers various stages of the drug development process, including peptide synthesis, small molecule design, drug dosage determination, and drug delivery effectiveness, showcasing the role of AI in streamlining and improving the drug discovery process.
Furthermore, the survey discusses the use of AI in structure and ligand-based virtual screening, quantitative structure-activity relationship (QSAR) modeling, drug repositioning, and predicting physicochemical properties and bioactivity of compounds. It also emphasizes the development of AI-based tools and software packages to assist in predicting drug-target interactions, evaluating drug toxicity, and identifying molecular pathways and polypharmacology. The survey acknowledges the potential of AI to revolutionize drug discovery and development, particularly in the context of neurodegenerative diseases, and highlights the challenges and opportunities associated with the use of AI in the pharmaceutical industry.The survey also provides a comprehensive list of tools and software used in drug discovery and development, including QSAR modeling, drug repurposing, prediction of drug-target interactions, prediction of toxicity, and prediction of physicochemical properties. These tools utilize machine learning, deep learning, neural networks, and other computational methods to integrate diverse chemical and biological data, predict drug-disease associations, and assess the mode of action and toxicity of compounds. The use of AI-based approaches in drug discovery and development is highlighted as a significant factor in predicting physicochemical properties.Overall, the survey contributes to the understanding of the significant role of AI and ML in advancing drug discovery and development, showcasing their potential to accelerate the process, improve efficiency, and revolutionize pharmaceutical research and healthcare. The survey also references various studies and developments in the field of AI and drug discovery, providing a comprehensive list of research papers and articles related to drug discovery, computational biology, and artificial intelligence in healthcare. These references cover a wide range of topics and applications within the field of pharmaceutical research and development, further emphasizing the impact and potential of AI in drug discovery and development.
References
- M. Vodanović, M. Subašić, D. Milošević, and I. S. Pavičin, "Artificial Intelligence in Medicine and Dentistry," Acta Stomatol Croat, vol. 57, no. 1, pp. 70–84, Mar. 2023. DOI: 10.15644/asc57/1/8. PMCID: PMC10243707.
- R. Gupta, D. Srivastava, M. Sahu, S. Tiwari, R. K. Ambasta, and P. Kumar, "Artificial intelligence to deep learning: machine intelligence approach for drug discovery," Springer Nature Switzerland AG, 2021.
- Y. You, X. Lai, Y. Pan, H. Zheng, J. Vera, S. Liu, S. Deng, and L. Zhang, "Artificial intelligence in cancer target identification and drug discovery," Signal Transduct Target Ther, vol. 7, p. 156, May 2022. DOI: 10.1038/s41392-022-00994-0. PMCID: PMC9090746. PMID: 35538061.
- H. Askr, E. Elgeldawi, H. A. Ella, Y. A. M. M. Elshaier, M. M. Gomaa, and A. E. Hassanien, "Deep learning in drug discovery: an integrative review and future challenges," Artif Intell Rev, vol. 56, no. 7, pp. 5975–6037, Nov. 2022. DOI: 10.1007/s10462-022-10306-1. PMCID: PMC9669545. PMID: 36415536.
- A. Blanco-González et al., "The Role of AI in Drug Discovery: Challenges, Opportunities, and Strategies," Pharmaceuticals, vol. 16, no. 6, p. 891, 2023. DOI: 10.3390/ph16060891.
- R. Qureshi et al., "AI in drug discovery and its clinical relevance," Heliyon, 2023. DOI: 10.1016/j.heliyon.2023.e17575.