Page 98 - OR-1-2

affect human tissues with a level of precision that reflects remain an area that needs further investigation and
individual variations in drug response. This innovation validation in the literature. Future work will be essential to
64
enables the detection of toxic reactions and potential side provide empirical evidence that quantifies these potential
effects much earlier in the drug development process, efficiency gains.
highlighting how different individuals may experience
varying degrees of toxicity. By simulating real human 4.3. AI-assisted organoids for drug screening
65
organ systems, organoids can identify harmful effects based AI-assisted organoids for drug screening highlight the
on unique genetic or environmental factors, improving the transformative role of AI in revolutionizing drug discovery.
70
personalization of drug safety evaluations. This approach By combining advanced AI algorithms with human-
66
not only enhances the precision of toxicity predictions but like 3D organoid models, this approach accelerates drug
also allows for more targeted screening of drug candidates testing and provides more accurate, personalized results.
71
to match individual susceptibility, reducing the risk of Organoids replicate the complexity of human organs,
adverse effects in clinical trials. offering a superior platform for drug testing compared
72
Modeling how individual variation in the human brain to traditional cell cultures. AI enhances this process by
affects disease susceptibility has long been a challenging analyzing vast amounts of data generated from organoid
task. Different human PSCs lines exhibit inconsistent experiments, quickly identifying patterns, and predicting
73
performance in in vitro models, which may be related to how different drugs interact with human tissues. This not
differences in reprogramming, epigenetic imprinting, or only speeds up the screening of potential drug candidates
sensitivity to culture conditions. Genetic variation plays a but also improves the precision of predictions, helping
67
74
crucial role in differential susceptibility to disease triggers. researchers select the most promising therapies. Through
However, although individual susceptibility can be detected AI’s ability to process and learn from complex data, this
in in vitro systems, exploring its underlying mechanisms has approach minimizes trial and error, reduces the need for
been hindered by the limited availability of experimental animal testing, and enables more effective, targeted drug
75
models. In July 2024, Antón-Bolaños et al. proposed a development. AI is the key to unlocking faster and more
69
68
highly reproducible multidonor human cortical organoid accurate drug discovery, making it a game-changer in the
model called Chimeroids. This model is generated by the pharmaceutical industry. 76
codevelopment of cells from individual donors within a Venous malformation (VM) is the most common type
single organoid, achieved through the reaggregation of cells of vascular malformation, with an incidence rate of 1 –
from multiple single-donor organoids at the neural stem or 2/10,000 and a prevalence of 1%. Pan et al. developed a
78
77
progenitor cell stage, resulting in a chimeric structure. The novel approach for constructing VM disease models and
team used Chimeroids to investigate individual differences screening therapeutic drugs using induced PSC (iPSC)
in susceptibility to neurotoxic drugs, such as ethanol and technology. The team manipulated cell cycle dynamics
the anticonvulsant valproic acid, which exhibit significant and employed the retinoic acid signaling pathway to
clinical phenotypic variability. Through techniques such generate induced venous endothelial cells (iVECs). By
as single-cell RNA sequencing and spatial transcriptomics, introducing the L914F mutation of the TIE2 gene into
the results indicated that the human genetic background the iPSC locus, they found that the mutated iVECs were
may be an important mediator of neurotoxin susceptibility. able to recapitulate the phenotypic features of VM after
Furthermore, the chimeric model provides a scalable both in vitro and in vivo transplantation. These features
system for high-throughput research into inter-individual included vessel dilation, abnormal smooth muscle cell
differences in brain development and disease processes coverage, increased cell proliferation, and enhanced
(Figure 9). The ability to perform large-scale drug response anti-apoptotic ability. Transcriptomic and proteomic
assays using Chimeroids holds the potential for clinically analyses revealed potential pathological mechanisms
stratifying patients into different treatment response groups associated with VM. In addition, using AI-based deep
based on data-driven insights. Over time, the accumulation learning prediction systems (DLEPS) and digital RNA
of large datasets may propel the development of universal sequencing (DRUG-seq) technologies for drug screening,
models for predicting drug efficacy before clinical trials. the team identified bosutinib as a potential therapeutic
Despite the immense potential of AI algorithms in drug agent. Bosutinib was found to reverse the VM phenotype
screening and improvements in efficiency, we currently by inhibiting endothelial-to-mesenchymal transition
lack direct, published validation studies with the specific (EndoMT), restoring cell function, and alleviating VM
quantitative data required to support these claims. Although symptoms. The iPSC-derived VM model and drug
AI-assisted methods are widely believed to enhance screening approach established in this study offer a
efficiency, reduce costs, and accelerate drug discovery, groundbreaking strategy for therapeutic research in
specific, data-driven comparisons with traditional methods vascular malformations (Figure 10).

Volume 1 Issue 2 (2025) 13 doi: 10.36922/OR025040005