The global pharmaceutical market is increasingly driven by antibody-based therapeutics and various antibody-integrated modalities. This growth is attributed to the unique advantages of antibodies as biologically derived proteins, which offer lower toxicity than small-molecule synthetic drugs while providing beneficial therapeutic properties such as immune modulation and in vivo recycling [1,2]. Recent breakthroughs in deep learning-based protein structure prediction [3,4] have further intensified interest in artificial intelligence (AI)-driven antibody drug discovery.

The development of effective antibody therapeutics requires precisely regulated target-binding affinity, mediated by complementarity-determining region (CDR) loops, along with fine-tuned physiological modulation upon target engagement [5]. Additionally, factors such as the immunological function of the Fc region, the structural flexibility of the hinge region, in vivo stability, and developability must be carefully optimized [6]. Achieving an antibody design that simultaneously meets these complex criteria remains a formidable challenge, yet continuous advances across multiple research domains are driving significant progress in this field [7].

Notably, antibody databases have played a critical role in fueling these developments. Despite the rapid expansion of sequence [8, 9, 10, 11] and structure [12, 13, 14] databases, experimentally validated data on binding affinity and other critical biological properties [15, 16, 17, 18] remain relatively scarce (see Table 1). Inspired by the success of AI-based protein structure prediction methods, AI approaches tailored specifically for antibodies are currently being developed. This review provides a concise analysis of the current state of the field, highlighting recent technological innovations and exploring future research directions.

The following sections will examine AI-based methodologies for protein-antibody complex structure prediction, antibody generation for protein targeting, and antibody property prediction, offering an overview of their principles, recent advancements, and implications for therapeutic development (see Figure 1).