AI-Integrated Molecular Pharmacovigilance in Immuno-Oncology: A Biochemical and Genomic Perspective

Background: Immunotherapy has revolutionized the management of cancer, whereas
immune-related adverse events(irAEs) are a significant issue when it comes to patient safety.
Traditional pharmacovigilance is very reactive and incapable of predicting these complex
and delayed toxicities. The advent of artificial intelligence (AI) has portended Onco-
Immuno-Pharmacovigilance, which combines oncology, immunology and
pharmacovigilance to predict safety in the form of future monitoring. At the molecular level,
immune-checkpoint inhibitors (anti-PD-1, anti-CTLA-4) interfere with immune tolerance
through changing cytokine signaling (IL-6, IFN- 7, TNF- 6) and T-cell metabolism. These
are regulated by genetic variation in HLA locus and CTLA4 and biochemical indicators of
early immune toxicity in ALT, LDH and IDO1.
Objective: To summarize the development of AI-based molecular pharmacovigilance, it is
important to highlight the need to combine biochemical pathways, genetic determinants and
multi-omics data into AI models to predict immunotherapy-related toxicities in a
mechanistic manner.
Methods: The databases such as PubMed, Scopus, WHO-VigiBase, and FDA-FAERS were
searched (2018-2025) on AI and immunotherapy and adverse drug reactions studies. The
preference was to include research based on the use of molecular biomarkers, groups of
cytokines, enzyme tests or the use of genomic variations associated with immune toxicity.
Results: Machine-learning models including genetic variations, cytokine, and enzyme assays
were more accurate in early warning of irAE and predicting hepatotoxicity or pneumonitis.
Mechanism validation of predictions made by AI derived prediction was mechanistically
DR.P. MAHESHWARI et al. J. APPL. BIOANAL
1166
validated through molecular analysis of HLA alleles, IL-6 and LDH/IDO1 activity. There
are still significant weaknesses such as bias, lack of interpretability and missing data.
Conclusion: Molecular pharmacovigilance is an AI that enhances pharmacovigilance beyond
a statistical surveillance approach to biochemical, genetic, and mechanistic accuracy and
permits the biologically meaningful and patient-specific safety surveillance by
interdisciplinary cooperation.