Unlocking the Future of Medicine: The Role of AI in Drug Discovery
Since 1995, Nature Medicine has been chronicling advances in medical technology and health care. To mark the journal’s 30th anniversary, each month an issue of critical importance for the future of medicine will be highlighted. Consistent with this theme, “Artificial Intelligence in Drug Discovery” in the January inaugural issue reviews advances in next-generation drug development, examining the transformative potential and application of artificial intelligence (AI) and quantum computing. AI, machine learning and quantum computing are now being utilized in various stages of drug development, from target identification, molecule synthesis, preclinical and clinical trials, and to post-market surveillance.
Target Identification
By leveraging vast biological databases, AI identifies promising drug targets such as small-molecule targets, proteins and nucleic acids. Past technologies such as affinity pull-down and whole-genome knockdown were slow and error-prone. AI addresses these limitations by accelerating the collection and analysis of large datasets within complex biological networks. The authors report that AI facilitates the identification of disease-related molecular patterns and causal relationships by constructing multi-omics data networks, thus facilitating the discovery of candidate drug targets.
De Novo Drug Design
AI greatly simplifies the design of new compounds, employing techniques like virtual screening to efficiently analyze vast libraries of potential drug candidates. For example, AI, particularly deep learning, has facilitated the automated identification of novel structures that meet specific requirements, bypassing traditional expertise.
Screening
AI streamlines the design of new compounds, employing technologies like virtual screening to efficiently sift through large libraries. For example, AI-based receptor–ligand docking models can predict ligand spatial transformations and directly generate complex atomic coordinates using algorithms.
Predicting Drug Properties
AI models are capable of forecasting critical drug properties such as absorption, metabolism, and toxicity. These predictions enable researchers to concentrate on the most promising drug candidates earlier in the development process.
Synthesis Planning and Drug Discovery
Chemical synthesis of new compounds in small-molecule discovery is a highly technical and extremely laborious task. Use of computer-aided synthesis planning (CASP) and automatic synthesis of organic compounds can accelerate the discovery process by removing laborious and repetitive tasks for chemists.
ADMET
ADMET (absorption–distribution–metabolism–excretion–toxicity) plays a critical role in determining drug efficacy and safety. AI techniques are supplementing wet-lab evaluations and can help reduce failures due to poor characteristics. These predictions enable researchers to concentrate on the most promising drug candidates earlier in the drug discovery process.
Drug Repurposing
AI is being used to identify new applications for existing medications by analyzing biological data patterns, expediting the development of novel treatments for various conditions.
Optimizing Clinical Trials
Clinical trials are expensive, time-consuming and inefficient. Drug developers are challenged with delays in clinical trial registration or struggle to find sufficient volunteers. AI has the potential to optimize trial design, streamline recruitment, and aid in biomarker discovery, thereby enhancing success rates and study efficiency. AI can also support continuous safety monitoring, adverse event reporting, and pharmacovigilance compliance, ensuring the long-term safety and efficacy of drugs.
What will medicine look like in 2055? Perhaps healthcare will have transitioned from a reactive model to one that is proactive, personalized and preventive. Advances in multiomics — integrating genomics, transcriptomics, proteomics and metabolomics — may enable precise prediction of individual disease risks. Continuous monitoring of biomarkers through wearable and implantable biosensors could allow early intervention, long before signs and symptoms manifest.
View referenced article
