Artificial Intelligence (AI)

The integration of artificial intelligence (AI) into biological research is revolutionising our ability to interpret complex biological systems and accelerate the development of next-generation diagnostic tools. In the context of infectious diseases, AI enables the identification of subtle and dynamic molecular patterns in host responses—patterns that are often undetectable using traditional analytical methods. This has paved the way for a shift toward host-based diagnostics, where biological signals from the individual, rather than direct pathogen detection, are used to identify infections.

One of the most impactful applications of AI in biology is the discovery of biomarkers—molecular indicators such as microRNAs, gene expression profiles, or protein signatures that reveal the presence, severity, or progression of disease. Unlike conventional statistical methods, AI algorithms can analyse high-dimensional, multi-omic datasets to uncover complex, non-linear relationships between biological features and clinical outcomes. These algorithms can be trained, validated, and optimised across diverse datasets, ensuring both robustness and generalisability. In the field of infectious disease diagnostics, AI-driven biomarker discovery enables the development of novel tests to fill gaps in our current diagnostic toolkit.

The Host Response Team has developed an AI-enabled biomarker discovery pipeline that processes raw sequencing data from NGS experiments, performs initial quality assessment checks, data preprocessing, and conventional differential expression analysis before moving to AI biomarker identification. We utilise several optimised models to identify the most predictive set of miRNAs for a given disease state, and then evaluate their classification performance, focussing on the minimum number of biomarkers needed for accurate disease prediction. The pipeline outputs putative biomarkers and a fine-tuned diagnostic AI model. The markers are then validated using the established regulatory frameworks for human or animal diagnostic tests and the model is incorporated into a user-interface (UI) that converts the lab output into actionable diagnostic results. This approach is at the centre of our innovative test for Johne’s Disease (JD) in cattle.

 

As data generation technologies advance and AI models continue to improve in precision and scalability, the development of host-response-based diagnostics will play an increasingly central role in global efforts to detect, monitor, and control infectious diseases.