Spectroscopic techniques such as infra-red, Raman, and mass spectrometry have long been used to identify chemical compounds and biological species, including bacteria and viruses, usually in specialised lab conditions with high performance instrumentation. Virus identification in realistic clinical/field environments, using low cost instrumentation, is appealing, as it can be widely deployed and so is very suitable for diagnosis, prevention and management in pandemics such as COVID-19. However, low cost instrumentation produces poorly-resolved spectra with added noise. Our recent work has investigated machine learning algorithms applied to spectra from low cost near infra-red (NIR) spectrometers to extract identifiable patterns from targets with complex backgrounds and limited experimental control/processing. Our latest study shows that it is possible to use the technique to accurately differentiate respiratory syncytial virus and Sendai virus in different media, and quantify their viral loads. We aim to develop a spectrometer-fronted, cloud-based system for in-situ SARS-CoV-2 detection with three deliveries. The system will record spectra from patient nasal samples in the field and return a positive/negative diagnosis within ~ 1 minute, based on model-driven analytics running on a cloud-based service. The detection model will be developed, trained and validated using spectra from the SARS-CoV-2 virus in (a) lysis buffer and (b) nasal aspirate simulant; the model will then be used to determine whether the virus is present in the sample using a 'subsumption' operation in the learning algorithm. The system will be validated in real environments in collaboration with our partners in Northern Ireland Regional Virology Lab (RVL).

Spectroscopic techniques such as infra-red, Raman, and mass spectrometry have long been used to identify chemical compounds and biological species, including bacteria and viruses, usually in specialised lab conditions with high performance instrumentation. Virus identification in realistic clinical/field environments, using low cost instrumentation, is appealing, as it can be widely deployed and so is very suitable for diagnosis, prevention and management in pandemics such as COVID-19. However, low cost instrumentation produces poorly-resolved spectra with added noise. Our recent work has investigated machine learning algorithms applied to spectra from low cost near infra-red (NIR) spectrometers to extract identifiable patterns from targets with complex backgrounds and limited experimental control/processing. Our latest study shows that it is possible to use the technique to accurately differentiate respiratory syncytial virus and Sendai virus in different media, and quantify their viral loads. We aim to develop a spectrometer-fronted, cloud-based system for in-situ SARS-CoV-2 detection with three deliveries. The system will record spectra from patient nasal samples in the field and return a positive/negative diagnosis within ~ 1 minute, based on model-driven analytics running on a cloud-based service. The detection model will be developed, trained and validated using spectra from the SARS-CoV-2 virus in (a) lysis buffer and (b) nasal aspirate simulant; the model will then be used to determine whether the virus is present in the sample using a 'subsumption' operation in the learning algorithm. The system will be validated in real environments in collaboration with our partners in Northern Ireland Regional Virology Lab (RVL).