AI-Powered UV Spectroscopy Detects Microbial Contamination in Record Time

A machine-learning-assisted ultraviolet (UV) absorbance spectroscopy method has been developed to detect microbial contamination in cell therapy product (CTP) manufacturing. Published in Nature, this breakthrough approach enables real-time, label-free detection, improving safety and efficiency.

Damaged area of DNA.
Study: Machine learning aided UV absorbance spectroscopy for microbial contamination in cell therapy products. Image Credit: Andrii Yalanskyi/Shutterstock.com

Download your PDF copy now!

The researchers used a one-class support vector machine (SVM) to analyze cell culture absorbance spectra, achieving rapid and accurate contamination detection with minimal sample preparation. When tested with seven microbial organisms, the method demonstrated a high detection sensitivity of 92.7 % (true positive rate) at low contamination levels (10 colony-forming units (CFUs)), performing comparably to standard methods.

Background

Advanced therapy medicinal products (ATMPs), including cell-based therapies, offer promising treatment options for severe and chronic diseases. However, their manufacturing process relies on nutrient-rich cell cultures that are highly susceptible to microbial contamination.

Traditional sterility testing methods, such as the United States Pharmacopeia (USP) <71> test, can take up to 14 days, require significant labor, and are prone to errors due to turbidity misinterpretation.

Rapid microbiological methods (RMMs) like BACTEC and flow cytometry offer faster detection but often require growth enrichment or staining, making them less ideal for ATMPs. Given the limited shelf life of CTPs, a rapid, non-invasive, and cost-effective contamination detection method is crucial.

This study introduced a machine-learning-enhanced UV absorbance spectroscopy method for real-time microbial contamination detection. By analyzing the absorbance spectra of mesenchymal stromal cell (MSC) cultures, the one-class SVM model identified contamination without labels or growth enrichment.

The method detected contamination at 10 CFUs within 21 hours—comparable to USP <71>—while requiring only minimal sample volume and preparation. This innovation provides a faster, scalable, and non-invasive solution to enhance safety in CTP manufacturing.

Machine-Learning-Aided UV Spectroscopy for Rapid Microbial Detection

To validate the method, researchers cultured MSCs from seven donors and used spent media to prepare both sterile and microbe-inoculated samples.

Seven microbial species, including Escherichia coli (E. coli) and Staphylococcus aureus (S. aureus), were introduced at low concentrations (10–100 CFUs). A commercial spectrometer measured UV absorbance spectra, and pre-processing was applied to correct for instrumentation drift and baseline variations.

A one-class SVM model, trained on sterile MSC samples, identified contamination by detecting spectral anomalies, particularly in the 237–300 nm range, where nicotinic acid (NA) and nicotinamide (NAM) exhibit distinct absorbance patterns.

The method achieved:

  • 92.7 % true positive rate and 77.7 % true negative rate, improving to 92 % after excluding samples with high NA levels.
  • Contamination detection at 10 CFUs within 21 hours, comparable to traditional sterility tests.
  • A label-free, non-invasive approach requiring minimal sample volume and preparation.

These findings highlight the method’s potential for real-time, at-line monitoring in CTP manufacturing, addressing key limitations of current sterility testing methods.

Findings and Analysis

The method successfully detected E. coli contamination at 10 CFUs within 21 hours, matching the performance of USP <71> and liquid chromatography-mass spectrometry (LC-MS)-based methods, though it was slightly slower than BACT/ALERT® 3D. In donor-specific tests, the model achieved 100 % true positive and true negative rates for E. coli.

When applied to other microbial species, including S. aureus, Candida albicans, and Cutibacterium acnes, the method maintained a 92.7 % true positive rate and 77.7 % true negative rate across six commercial donors, consistently detecting contamination at 10 CFUs.

To assess robustness, researchers analyzed donor variability. Donors A and B showed the highest prediction accuracy, while Donor F exhibited higher NA levels, leading to false positives. LC-MS analysis confirmed that NA concentration differences influenced model predictions. Principal component analysis (PCA) further revealed that spectral differences between NA and NAM at around 268–270 nm were key to contamination detection.

By eliminating the need for growth enrichment or complex sample preparation, this label-free, non-invasive method presents a promising, cost-effective solution for real-time sterility testing in cell therapy manufacturing.

Conclusion

This study demonstrates that machine-learning-aided UV absorbance spectroscopy enables rapid, non-invasive microbial contamination detection in CTP manufacturing. By analyzing MSC culture absorbance spectra, the one-class SVM model detected contamination at 10 CFUs within 21 hours, achieving a 92.7 % true positive rate and 77.7 % true negative rate across six donors. The method leverages spectral differences between NA and NAM, allowing real-time, label-free detection without requiring growth enrichment or complex preparation.

Future research should focus on expanding validation across a broader range of microbial species, stem cell types, and donor demographics. Integrating this approach with automated sampling systems could enable continuous, real-time contamination monitoring in CTP manufacturing, further enhancing safety and reducing risks.

Journal Reference

Pandi Chelvam et al., 2025. Machine learning aided UV absorbance spectroscopy for microbial contamination in cell therapy products. Scientific Reports15(1). DOI:10.1038/s41598-024-83114-y. https://www.nature.com/articles/s41598-024-83114-y

Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.

Citations

Please use one of the following formats to cite this article in your essay, paper or report:

  • APA

    Nandi, Soham. (2025, March 14). AI-Powered UV Spectroscopy Detects Microbial Contamination in Record Time. AZoRobotics. Retrieved on April 30, 2025 from https://www.azorobotics.com/News.aspx?newsID=15795.

  • MLA

    Nandi, Soham. "AI-Powered UV Spectroscopy Detects Microbial Contamination in Record Time". AZoRobotics. 30 April 2025. <https://www.azorobotics.com/News.aspx?newsID=15795>.

  • Chicago

    Nandi, Soham. "AI-Powered UV Spectroscopy Detects Microbial Contamination in Record Time". AZoRobotics. https://www.azorobotics.com/News.aspx?newsID=15795. (accessed April 30, 2025).

  • Harvard

    Nandi, Soham. 2025. AI-Powered UV Spectroscopy Detects Microbial Contamination in Record Time. AZoRobotics, viewed 30 April 2025, https://www.azorobotics.com/News.aspx?newsID=15795.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this news story?

Leave your feedback
Your comment type
Submit

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.

Read the full Terms & Conditions.