Artificial intelligence (AI) is supporting laboratory diagnostics across various fields, from research and drug discovery, to screening and monitoring disease outbreaks. But the future remains uncertain as laboratory professionals hesitate to adapt and collaborate towards wider adoption.
Here, we hear from three experts about the risks and opportunities of AI applications in infectious disease diagnostics.
Where is AI already having the greatest impact on infectious disease diagnostics?
Moritz Kraemer: AI and machine learning models can integrate and harmonize large volumes of spatially resolved data, helping researchers identify areas at increased risk of disease outbreaks and improve the accuracy of outbreak prediction and forecasting models.
Bram Lestrade: AI is increasingly being incorporated into diagnostic platforms for pathogen identification and antimicrobial susceptibility testing (AST). Commercial systems already use AI to support data analysis and result interpretation.
AI is also being applied to microscopic image analysis, enhanced MALDI-TOF identification, and, to a lesser extent, MALDI-TOF–based susceptibility testing. While in-house models can be effective, broader implementation is often limited by validation and regulatory requirements. Commercial platforms generally have an advantage because they are developed within established compliance frameworks.
Stephan Sieber: One of the biggest challenges in infectious disease diagnostics is the rapid identification of bacterial infections. Traditional culture-based methods can take several days, delaying treatment decisions.
Our group is developing small-molecule probes that bind to specific bacterial enzymes or proteins and generate distinctive fluorescent signals. These signatures may enable rapid identification of bacterial species, strains, or substrains directly from patient samples without the need for culture. Although still under development, this approach illustrates how AI-enabled and molecular technologies could support faster, more targeted diagnostics.
What opportunities does AI offer for improving accuracy and turnaround times in laboratory testing?
BL: AI has the potential to accelerate both pathogen identification and AST. Some platforms already provide results within hours for selected pathogens and sample types, although performance still varies and validation remains essential.
Beyond testing itself, AI may improve laboratory workflows through automated sample prioritization and result review. For example, AI could help determine whether pathogen identification is consistent with Gram stain findings and susceptibility results, or flag potential reporting errors.
AI-assisted image analysis is another promising area. Automated interpretation of Gram stains and microscopy-based detection of parasites and fungi could support faster reporting while reducing interobserver variability.
What risks or limitations should laboratories consider when adopting AI tools?
SS: One of the biggest risks is overestimating what AI can currently achieve. AI can accelerate research and support decision-making, but it does not replace experimental validation. Results still need to be tested, verified, and interpreted critically.
BL: AI models can also perform less reliably when applied to patient populations, datasets, or clinical settings that differ from those used during development. Laboratories should view AI as a decision-support tool rather than a substitute for expert judgment.
Validation should be continuous, as laboratory procedures, guidelines, and patient populations change over time.
Can AI enable more proactive surveillance, and what role do laboratory data play in this?
MK: AI can support surveillance systems by enabling earlier outbreak detection, monitoring transmission dynamics, and characterizing emerging pathogens through genomic analysis.
BL: Faster and more accurate laboratory data can support earlier public health interventions, helping to interrupt transmission and limit disease spread. Because laboratories generate much of the data used to identify and track pathogens, they remain central to surveillance and outbreak response efforts.
What regulatory or validation hurdles must be overcome for wider adoption in diagnostics?
SS: Strong validation remains essential. Many AI models perform well in development datasets, but demonstrating reliable performance in real-world settings requires rigorous testing in independent populations.
BL: Regulatory requirements, including the EU In Vitro Diagnostic Regulation (IVDR), present additional challenges. Successful implementation will require ongoing validation, regulatory oversight, and collaboration across institutions.
AI adoption also depends on multidisciplinary teams that bring together expertise in clinical microbiology, infectious diseases, data science, implementation, and regulation. No single discipline can address these challenges alone.
Looking ahead, how do you see AI reshaping infectious disease diagnostics in the next decade?
MK: We are likely to see rapid advances in AI models and data availability, creating opportunities for more sophisticated diagnostics and surveillance systems. However, these tools must continue to be evaluated against established methods, and access should extend beyond well-resourced institutions.
BL: Faster diagnostic testing is likely to be one of the most important developments. More rapid pathogen identification and susceptibility testing could enable earlier targeted therapy, reduce unnecessary use of broad-spectrum antibiotics, improve patient outcomes, and help combat antimicrobial resistance.
AI may also support more detailed pathogen characterization, improved treatment decisions, and stronger infection surveillance. However, these benefits will depend on ongoing validation and meaningful human oversight.
SS: The ability to identify pathogens more quickly and accurately will be critical. Earlier diagnosis allows clinicians to move from empirical treatment to targeted therapy sooner, improving care while reducing selective pressure for resistance.
Ultimately, AI's greatest impact may be its ability to accelerate both diagnostics and drug development, helping clinicians and researchers make faster, more informed decisions.
Newsletters
Receive the latest pathologist news, personalities, education, and career development – weekly to your inbox.
