Artificial intelligence in the development of small nucleic acid therapeutics: toward smarter and safer medicines

Small nucleic acid therapeutics, including antisense oligonucleotides (ASOs), small interfering (si)RNAs, miRNAs, and aptamers, modulate gene expression through complementary sequence recognition. Unlike traditional drugs, they can target previously inaccessible pathways and are showing promise for treating genetic, infectious, and degenerative diseases. Yet, their clinical translation is often constrained by poor stability and unpredictable pharmacokinetics. In this review, we focus on the main classes of small nucleic acid drugs, their mechanisms of action, and the chemical modifications that enhance their pharmacological properties. Importantly, we provide a critical comparison of different artificial intelligence (AI)-based methodologies for the design and optimization of oligonucleotide therapeutics, discuss their limitations (including data scarcity, off-target predictions, and clinical translation barriers), and outline concrete directions for future improvements.