Page 36 - AIH-2-2

Artificial Intelligence in Health New drug discovery in the AI era

and prioritize compounds for wet-laboratory profiling next-generation drug discovery. In silico absorption,
4
accordingly. As a result, the chemical synthesis of novel distribution, metabolism, and excretion (ADME)
moieties for wet-laboratory evaluation has been reduced screening, combined with cost-effective and less labor-
to one-tenth of the previous workload. AI predictive tools intensive in vitro studies, is being adopted in early drug
are also helping researchers identify potentially unsuitable discovery to selectively eliminate compounds with poor
molecules early in the pre-clinical stage, allowing them ADME and PK attributes. 5
to “fail fast” and save both time and resources. This Innovative approaches are being used to design targeted
revolutionary AI approach has not only impacted the chemical libraries or fine-tune the ADME profile of NCEs
discovery and pre-clinical development of new drugs but transitioning into lead optimization to reduce late-stage
has also refined clinical trials through improved patient attrition rates. In addition, developing in vitro model
6
selection and stratification, generation of early safety and systems that resemble or closely mimic human tissues or
tolerability warning signals, and real-time collation of organs to predict acute drug toxicity or establish a PK/PD
multicentric clinical trial data, thereby increasing success relationship to reduce clinical-stage failures is becoming a
rates. AI has further facilitated the generation of clinical trend. The USFDA recently accepted pre-clinical efficacy
7,8
trial procedures and reports, helping clinicians draw results from human organ-on-a-chip models and approved
meaningful insights from the vast data produced during the clinical trial IND application for the mAb sutimlimab,
multicentric trials. This mini-review briefly discusses the manufactured by Sanofi.
conventional drug discovery and pre-clinical development
process, setbacks associated with traditional methods, 2.1. Drug safety and toxicity
and how AI is bridging these gaps to accelerate new drug
development. Furthermore, it explores the challenges and Drug safety and toxicity evaluation, spanning both
limitations associated with the implementation of AI in pre-clinical and clinical trials, is a crucial step in drug
discovery and development. Its aim is to identify and
both pre-clinical and clinical settings.
assess any potential side effects or adverse responses to
2. Drug discovery and development the drug. Pre-clinical safety evaluations primarily rely on
animal testing in rodent and non-rodent species, with
Drug discovery and development is a challenging, variations in duration, design, and objectives. These studies
lengthy, and costly process. The time taken for a drug include general toxicity, reproductive and developmental
to move from the wet-laboratory stage to market is toxicity, carcinogenicity, immunotoxicity, and functional
approximately 10 – 12 years, with costs ranging from $161 evaluations of key organ systems, such as the respiratory,
million to $4.54 billion. 1 central nervous, and cardiovascular systems. Identification
Despite the investment of billions of dollars, significant of possible toxicity in humans, characterization of the
efforts, and resources, nearly nine out of 10 potential drug toxicity (morphology, dose-response, reversibility, etc.),
2
candidates fail in clinical trials, and only one progresses and assessment of whether it can be effectively monitored
from bench to bedside. The Center for Drug Evaluation and and managed in human clinical trials are the main
Research (CDER-USFDA) approved 50 new drugs in 2021, objectives of pre-clinical drug safety assessment research.
37 in 2022, 55 in 2023, and 22 (as of this writing) in 2024. Pre-clinical studies also produce endpoints commonly
For NCEs entering first-in-human trials, the failure rate used to assess health and disease in humans, such as serum
remains high: around 80% in the cardiovascular segment, biochemistry, hematology, urinalysis, histopathology, and
84% in arthritis pain and infectious diseases, and 92 – 95% vital organ function evaluation.
in oncology and central nervous system therapeutics. Some Conventionally, NCEs progress through drug discovery
reasons for the high attrition rate in clinical trials include stages until crucial data reveal a low safety margin
off-target toxic side effects/unmanageable toxicity, poor (therapeutic index), suboptimal efficacy at clinically
PK properties, and suboptimal clinical efficacy. Figure 1 relevant doses, or an undesirable PK profile. If the pre-
3
shows the various stages of classical drug discovery and the clinical data are unconvincing, further development of
pre-clinical development flow. the NCE is halted, and sometimes, the entire program for

More focused efforts are now being made to develop that target is abandoned. This causes significant setbacks,
methods and approaches that accelerate the drug resulting in considerable losses of time, resources, and
discovery process, reduce research and development costs, money. For selected candidates, drug safety and tolerability
and increase the success rate of clinical candidates. Assay present major clinical challenges, with safety concerns
miniaturization technologies and the availability of highly accounting for 35% of Phase I failures and 28% of Phase
selective, sensitive analytical instruments have shaped II failures.

Volume 2 Issue 2 (2025) 30 doi: 10.36922/aih.4423