KAIST and Neogenlogic Unveil AI Platform for Personalized Cancer Vaccines

On Jan. 2, 2026, a joint research team from the Korea Advanced Institute of Science and Technology and Neogenlogic unveiled an artificial intelligence platform that analyzes a patient’s tumor genome to predict neoantigens, the mutation-derived protein fragments unique to an individual’s cancer. The partners say the system is intended to speed identification of the molecular targets that personalized cancer vaccines aim to train the immune system to recognize.

Personalized cancer vaccines rely on neoantigens because these mutated peptides are present only on tumor cells and can, in principle, provoke a focused T cell attack while sparing healthy tissue. Identifying the most promising neoantigens has proved a major bottleneck: it requires deep sequencing of tumor and normal DNA, bioinformatic filtering of thousands of mutations, and assessment of a peptide’s ability to be presented by a patient’s human leukocyte antigen molecules and to elicit an immune response. The KAIST and Neogenlogic platform addresses that pipeline with computational models that prioritize candidate neoantigens from genomic data, according to the announcement.

The developers frame the platform as a tool to compress the time from biopsy to vaccine design and to improve the accuracy of antigen selection. Faster, more reliable prediction could make personalized vaccines feasible for a wider range of patients and tumor types, particularly those with a modest mutational burden where precise selection matters most. Computational acceleration also matters commercially: reducing time and cost in the design phase can make individualized therapies easier to manufacture and scale.

Technical and clinical hurdles remain. Predicting which neoantigens will bind major histocompatibility complex molecules and then be recognized by T cells is a probabilistic challenge; many candidates selected in silico do not generate effective immune responses in patients. Tumor heterogeneity and immune-suppressive microenvironments can blunt vaccine efficacy. The platform’s utility will depend on laboratory validation of predicted neoantigens, integration with rapid manufacturing pipelines for peptide or mRNA vaccines, and ultimately demonstration of safety and benefit in early-stage clinical trials.

The announcement did not disclose detailed performance metrics or clinical results. Absent such data, oncologists and immunologists will evaluate the platform on its ability to reproduce known immunogenic neoantigens, its false positive rate, and how well it generalizes across diverse patient HLA types and tumor histologies. Regulators will also scrutinize data provenance, genomic privacy protections, and manufacturing quality controls if the platform proceeds toward clinical use.

Beyond efficacy, the technology raises questions about access and equity. Personalized vaccines require genomic sequencing, sophisticated analytics, and bespoke manufacturing capacity that are concentrated in academic centers and commercial hubs. If AI-driven tools accelerate development but remain costly or centralized, disparities in access to personalized immunotherapies may persist.

KAIST and Neogenlogic positioned the platform as a step toward more nimble personalized immunotherapy programs. The next phases will be laboratory confirmation of predicted neoantigens and clinical testing to determine whether AI-guided selection translates into meaningful patient benefit. The broader field will be watching closely to see whether computational advances can finally convert genomic insight into durable, individualized cancer treatments.