Traditional drug design and development is a long, complex, and resource-intensive process with an estimated average cost of US$2.6 billion and 12 years needed to discover new drugs. The advancements in artificial intelligence (AI), machine learning (ML), deep learning (DL), and access to extensive biological data fuel the accurate identification of patterns and models for drug discovery. AI and ML enable drug discovery and development at lower investments of time, workforce, and costs. Below, we discuss how artificial intelligence (AI) can help in lowering the cost of drug discovery and development.
Artificial Intelligence (AI) is a broad field that encompasses various subfields, including machine learning (ML), which consists of supervised, unsupervised, and reinforcement learning. Deep learning (DL) is a type of ML that utilizes neural networks to simulate the functions of the human brain. In the context of drug design and discovery, some common AI algorithms employed include artificial neural networks, deep neural networks, and support vector machines (SVMs).
Applications of AI
Computer-aided drug design (CADD) with integrated AI algorithms can address the challenges of traditional drug design and development. AI has been implemented from peptide synthesis to molecule design, virtual screening (structure- and ligand-based) to molecular docking, quantitative structure-activity relationship (QSAR) to drug repositioning, protein misfolding to protein-protein interactions, and molecular pathway identification to polypharmacology. In drug discovery, ML and DL algorithms have been applied in toxicity prediction, drug monitoring and release, pharmacophore modeling, drug repositioning, and physiochemical activity. DL is used in all stages of drug development, including drug-target interactions, drug-drug similarity interactions, drug sensitivity, and responsiveness, and drug-side effect predictions. Deep reinforcement learning has been successfully used to develop novel de novo drug designs.
Computer-Aided Drug Design (CADD)
CADD is used to discover, develop, and analyze drug candidates and active molecules, accelerate drug discovery, reduce failure rates, and reduce costs. CADD reduces drug development costs and time, with conservative estimates suggesting AI pipelines incur less than 1/3 of the prevailing time and cost. Applications of big data and AI in CADD enable the accurate and comprehensive analysis of big biomedical data and the development of predictive models. Computational methods analyze biomedical entities for target identification effectively and hit hunting and to gain safety, efficacy and avoid toxicity for drug development.
Predicting Binding Affinity
Accurately predicting the binding affinity and interaction strength between the ligand and the target protein is crucial in drug discovery. Compared to in vitro experiments, computational approaches to predict binding affinity significantly reduces drug discovery time and cost.
Drug Toxicity Prediction
In drug development, AI has reduced time, cost, attrition rates, and the need for human resources. The in vivo test for drug toxicity increases drug discovery costs, while computational methods can predict toxicity at a low cost and high efficiency.
In vitro and in vivo experiments are time-consuming and expensive. The cost-effective and time-efficient AI techniques have been effectively applied to predicting drug bioactivities, e.g., anti-cancer, antiviral, and antibacterial activities. Stokes et al. (2020) model for predicting antibacterial activity: http://chemprop.csail.mit.edu/.
Drug Synergism and Antagonism
While experimentally investigating drug combination effects of drugs is costly and time-consuming, advancements in AI offer a lower cost and more efficient option. Preuer et al. (2018) proposed a DL-based model for predicting synergism of anti-cancer drugs – the web server and source code provided at www.bioinf.jku.at/software/DeepSynergy and https://github.com/KristinaPreuer/DeepSynergy.
Virtual Screening (VS)
Classified as structure-based (SBVS) and ligand-based (LBVS), accelerates drug discovery and shortens the number of compounds to be tested in the wet lab. Additionally, VS plays a vital role in drug repurposing or repositioning and quickly optimizing the drug candidates. VS saves time and costs for drug discovery and efficiently obtains various molecule scaffolds. Example: CNS drug discovery is incredibly challenging, exacerbated by long timelines and high attrition rates. AI/ML applied in screening and lead discovery through VS enables cheaper and faster screening of larger compound libraries. Additionally, applying ML methods in SBVS increases the robustness and accuracy of scoring functions (SFs), conformational sampling, and ranking.
DL and Cancer Biomarker Discoveries
Large omics datasets have enabled data-driven biomarker discoveries. ML/DL approaches for performance optimization across datasets produce robust ensemble-learning prediction models, which also have a role in precision medicine.
Success Rates of Clinical Trials
The use of Artificial Intelligence (AI) in drug discovery and development has the potential to revolutionize the entire industry, but while it may still have a long way to go, it’s already starting to show significant improvements. One area where AI has already made a noticeable impact is in the success rates of clinical trials. The application of AI reduces clinical trial costs by, e.g., analyzing toxicity and side effects. For example, IBM Watson developed a system that uses AI to match clinical trial patients and track their progress. Although the application of AI in drug discovery and development is still in its infancy, the benefits it can bring to clinical trials are already beginning to show.
Clinical Trial Design, Conduct, and Analysis
As the pharmaceutical industry continues to grow and evolve, so do the challenges that come with drug discovery and development. With the increasing complexity of data (volume, variety, and velocity), and engagement of contract research organizations (CROs), sponsor oversight of site performance and clinical trial data has become challenging, time-consuming, and costly. Across all study phases, trial site monitoring is among the top three clinical trial expenditure cost drivers (9–14% of total cost). Fortunately, the introduction of artificial intelligence (AI) has brought about positive changes in clinical trial design, conduct, and analysis.
While there is still a long way to go in terms of fully integrating AI into drug discovery and development, the benefits it has already brought to clinical trial management are significant. By enabling remote monitoring, improving patient matching, and providing real-time insights, AI has the potential to revolutionize the way clinical trials are conducted. As AI technology continues to advance, we can expect to see even more innovative solutions that will further streamline and improve the drug development process.
Stay Connected with Vial CRO
The Vial Technology Platform leverages connected systems and intuitive design to run global trials efficiently at scale. Vial CRO’s eClinical Suite delivers high-quality data through connected data capture and review. Vial CRO replaces paper source in trials — driving a significant change in trial workflows. Driven by process automatization, Vial CRO’s Site Startup app enables lightning-fast onboarding. Sites can be activated in as few as 30 days. Vial CRO’s Patient Recruitment Platform recruits patients across 15+ channels and converts them to randomizations — radically reducing enrollment periods. For more info on how Vial CRO can help you manage your next clinical trial and reduce costs, visit us at vial.com or connect with a team member today!