Strengthening Australia’s Biosecurity Using AI-enabled Diagnostics

May 11th, 2023

Effective diagnostics are key to managing infectious disease outbreaks. We're using AI to underpin the next generation of biosecurity diagnostics.

The challenge

Infectious diseases pose a significant threat to global health and agriculture, and Australia is no exception. With the emergence and spread of new and deadly pathogens, it is crucial that we have effective diagnostic tests in place to identify and respond to these threats. However, the current diagnostic tests for infectious diseases have several challenges that make them less than ideal for Australian biosecurity.

Current diagnostic tests can often lack sensitivity and specificity, meaning that they may not accurately detect the presence of a pathogen or may produce false-positive results. This can lead to unnecessary treatment and a potentially dangerous false sense of security. Some tests detect the disease too late, after which treatment options are limited and the disease may have spread to other individuals. In some cases, there is simply no diagnostic test available. 

 

Limitations of some current diagnostic tests

 

These challenges highlight the need for new and improved diagnostic tests that are sensitive, specific, rapid, and adaptable to emerging threats. Diagnostics has been highlighted as one of the National Biosecurity Priorities by the Department of Agriculture, Fisheries, and Forestry. By investing in new technology and innovative solutions, we can improve our ability to detect and respond to infectious diseases and protect Australia’s biosecurity.       

 

Credit: National Biosecurity Co​mmittee (NBC) Department of Agriculture, Fisheries and Forestry (DAFF)

 

A new approach

One promising solution is the use of microRNA (miRNA) biomarkers [1]. MicroRNAs are small RNA molecules that regulate gene expression, and recent research has shown that changes in their expression can be indicative of the presence of certain pathogens and diseases.

MicroRNA-based diagnostic tests have several advantages that make them particularly useful for strengthening Australia’s biosecurity.

 

Tribolet (2020). Frontiers in Microbiology

 

They:

  • Respond quickly to stimuli, including infections, and can produce detectable changes before symptoms occur or the pathogen itself can be found [2], enabling earlier disease detection,
  • Are found in all biological fluids, allowing systemic detection of localised disease,
  • Show remarkable stability and are easily measured using standard molecular biology techniques,
  • Can differentiate between active infections and vaccinated animals, a pitfall of the commonly used antibody tests, and
  • Can be used for a range of applications, including predicting disease severity and outcomes, and monitoring the treatment efficacy. This versatility makes them an invaluable tool for strengthening Australia’s biosecurity across multiple sectors.

In recent years, there has been significant progress in the development of microRNA-based diagnostic tests for a range of infectious diseases [1, 3-5]. As this technology continues to evolve, it has the potential to revolutionise the way we detect and respond to infectious diseases, ultimately strengthening Australia’s biosecurity and protecting the health and wellbeing of our population and that of the plants and animals we rely on.

 

AI-powered biomarker discovery (WP1)

Artificial intelligence (AI) is a rapidly developing field that has shown promise in improving the identification of miRNA biomarkers for infectious diseases. By using machine learning algorithms to analyse large datasets of miRNA expression patterns, AI can identify unique patterns and signatures that are indicative of the presence of a pathogen.

One of the key advantages of using AI for miRNA biomarker identification is the ability to analyse large amounts of data in a short period of time. Traditional methods of identifying miRNA biomarkers are often time-consuming and require significant human input, making them impractical for large-scale analysis. AI can process data at a much faster rate, allowing for the identification of new and previously unknown biomarkers. Furthermore, current univariate statistical approaches (such as differential expression) assess each miRNA separately. As no biological process occurs in a vacuum, it is important to build multivariate patterns to identify biologically relevant changes – an approach where AI excels.

The use of AI to formulate a predictive diagnostic AI model using miRNA expression will underpin the next generation of biosecurity diagnostics. This work package will construct a semi-automated data analysis pipeline to allow rapid, transparent, and reproducible identification of miRNA biomarkers and optimisation of predictive AI models. This pipeline will be constructed in conjunction with our wet-lab scientists and domain experts, ensuring that it is guided by biological principles and results in targeted molecular assays.

 

Biomarker discovery platform approach

 

Overall, the use of AI for miRNA biomarker identification has the potential to significantly improve our ability to detect and respond to infectious diseases that can affect Australian biosecurity. By investing in this technology and developing new and innovative solutions, we can protect the health and wellbeing of our human, animal, and plant populations.

 

Diagnostic AI Interface (WP2)

Incorporating a diagnostic AI model into a human-centered user interface is essential for translating molecular diagnostic test outputs into actionable information. A user interface that is intuitive and user-friendly is crucial for ensuring that the results of diagnostic tests are effectively communicated to stakeholders and decision-makers.

The first step in creating a human-centered user interface is to identify the needs and requirements of the end-users. This involves understanding the knowledge and expertise of different stakeholders, including laboratory technicians, veterinarians, and biosecurity officials. By gathering feedback and input from these stakeholders, we can ensure that the user interface is tailored to their needs and effectively supports their decision-making processes.

Once the needs of the end-users have been identified, the diagnostic AI model (from WP1) can be integrated into the user interface. The AI model can analyse the results of diagnostic tests and provide real-time feedback on the presence of a disease. This information can be presented in a user-friendly format, such as a graphical representation, that allows stakeholders to quickly and easily understand the results of the diagnostic test.

We plan on deploying this diagnostic interface (in conjunction with the molecular assay) by hosting the core AI model on a CSIRO server to enable real-time optimisation and rapid roll out of future biomarker assays.

 

Planned diagnostic AI interface

 

Use Case 1 – Johne’s Disease in cattle (WP1)

Johne’s Disease (JD) is a fatal wasting disease caused by Mycobacteria avium subspecies paratuberculosis (MAP). It mainly affects ruminants, causing significant meat and milk production losses, lowering the price of goods, and restricting export opportunities. The current diagnostic tests for JD in cattle have several shortcomings that limit their effectiveness in detecting and managing the disease. One major limitation is the inability to detect the disease in its early stages due to the long incubation period, which can last for several years, during which infected animals may not show any clinical signs or shed detectable levels of the bacteria.

The gold standard diagnostic test for JD is a culture assay that requires 3-6 months for a result, and which displays roughly 50% sensitivity, leading to large numbers of false negatives.  Due to this, many diagnostic laboratories require multiple samples to be taken from an animal over an extended period of time, which can be time-consuming and costly.

Overall, improving diagnostic tests for Johne’s Disease in cattle is critical for the health and welfare of animals, and the economic sustainability of the cattle industry. Advances in diagnostic technology, such as the use of microRNA biomarkers and AI-enabled diagnostic models, hold promise for the development of more effective and efficient diagnostic tests for this important disease.

Photo by Richard Gatley on Unsplash

Use Case 2 – Microbial pathogens of plants (WP3)

Many endemic and exotic microbial pathogens cause plant diseases that negatively impact Australian plant-based agriculture. Effective identification of these pathogens is vital so that control measures are deployed in a timely manner. Of concern however is the lack of innovative diagnostic technologies in the biosecurity system to aid in the effective identification of these many plant pathogens. This is particularly true for front line plant health diagnostic laboratories who largely depend on older established diagnostic technologies. An opportunity exists therefore to develop novel diagnostic platforms and technologies that can enhance and improve Australia’s capacity to identify pathogens of biosecurity concern.

Research through AI4M, centred on plant host biomarkers, will help Australian plant-based industries identify some of their most important plant pathogens at the pre-border, border and post-border. Pathogen-specific plant host biomarkers could be used by Federal biosecurity and post-entry quarantine laboratories to screen incoming commodities and plant germ plasm for exotic national priority pathogens. Post-border, this AI-enabled technology could be utilised by State plant diagnostic laboratories to identify selected high priority pathogens of concern to plant industries. Furthermore, this technology is potentially platform agnostic so could be incorporated into in-field diagnostic devices used by farmers, agronomists and fee-for-service diagnostic laboratories.

Circulating noncoding molecules are targeted as disease biomarkers in animal and human systems. However, they have not specifically been utilised in plant systems to identify pathogens. CSIRO has developed unique analytics pipelines to identify informative host biomarkers associated with pathogen infection. Our value proposition is to take advantage of this pipeline, through AI4M, and apply it to the plant health system to identify plant biomarkers in response to priority plant pathogens of concern to Australian plant-based industries. This technology is particularly suited to many plant diseases with long symptom lag phases and could provide early disease confirmation for timely downstream management to save revenue for growers. Using a biomarker discovery and validation protocol designed by CSIRO, we believe we can adapt this for applicability to plant health and plant biosecurity in an Australian context. This will allow us to identify novel plant biomarkers specific for important exotic and endemic plant diseases of particular importance to the Australian horticultural and grains industries and those pathogens that post a threat to Australia’s native flora. Should this succeed, it would deliver a novel diagnostic platform and biomarkers into the Australian plant health sector. This work package will focus on identifying the best use case pathosystem most applicable to the Australian horticultural or grains industry.

Wheat rust. Credit: Melania Figueroa, CSIRO

Project Team

Ryan Farr, Gavin Hunter, Carlos Rodrigues, Cameron Stewart, Christopher Cowled, Shannon Dillon, Rob Dunne, James Doecke

This work is generously supported through Artificial Intelligence for Missions (AI4M) and Catalysing Australia’s Biosecurity (CAB).

 

Contact

Dr. Ryan Farr

Research Scientist

  • Ryan is the overall project leader as well as the leader for work packages 1 and 2. He is a wet-lab turned dry-lab scientist, with expertise in molecular biology, virology, bioinformatics and machine learning.

Dr. Gavin Hunter

Team Leader

  • Gavin is the lead for work package 3, centred on plant diagnostics. He is a molecular plant pathologist and mycologist and current team leader of the Molecular Plant Diagnostics team.

References

  1. Tribolet, L., et al., MicroRNA Biomarkers for Infectious Diseases: From Basic Research to Biosensing. Front Microbiol, 2020. 11: p. 1197.
  2. Stewart, C.R., et al., Promotion of Hendra virus replication by microRNA 146a. J Virol, 2013. 87(7): p. 3782-91.
  3. Farr, R.J., et al., Altered microRNA expression in COVID-19 patients enables identification of SARS-CoV-2 infection. PLoS Pathog, 2021. 17(7): p. e1009759.
  4. Poore, G.D., et al., A miRNA Host Response Signature Accurately Discriminates Acute Respiratory Infection Etiologies. Front Microbiol, 2018. 9: p. 2957.
  5. Biswas, S., et al., Development and validation of plasma miRNA biomarker signature panel for the detection of early HIV-1 infection. EBioMedicine, 2019. 43: p. 307-316.