Periodic Reporting for period 1 - SepsISensoR (Sepsis Diagnosis via Integrated Breath Sensing System with Change-Point Detection for Real-Time Point-of-Care)
Période du rapport: 2022-10-03 au 2025-11-02
Every 2.8 seconds someone dies from sepsis: 11 million people annually, out of which children under 5 years old. Septic shock is the potentially fatal body’s dysregulated response to pathogens that spread through blood circulation, including viruses like SARS-CoV-2. Half of sepsis cases happen in ICUs with 42% morality, with costs of €30 000 per case and €20 billion per year. Current diagnosis protocols rely on observation of the clinical symptoms to initiate regular monitoring of patients’ vital signs. Blood analyses and other tests identify the source of infection within 2-5 days, during which broad-spectrum antibiotics are administered, contributing to antibiotic resistance. Every hour of delay increases the mortality rate by 5-10%. Hence, early diagnosis of the infection source is a major step towards treatment. Pre-clinical and commercial point-of-care devices reduce the analysis time to few hours but still rely on the manifestation of clinical symptoms and invasive blood-based assays. SepsISensoR will advance the sepsis diagnosis protocol by non-invasive real-time monitoring of ICU patients’ breath to detect pre-symptomatic signs of sepsis based on transient changes of gas biomarkers in early sepsis stages. This will be achieved by: (a) integrating commercial gas sensors with fabricated preconcentrators on a single chip for high sensitivity, efficient and scalable multiple gas sensing; (b) using on-line change-point detection (CPD) on the breath signal to identify temporal variations of single- and multi-gas concentration; (c) validating the system with gases released from in vitro and in vivo models of sepsis. SepsISensoR will go beyond the state-of-the-art by reducing diagnosis time, and in turn reducing time in ICUs, cost, and sepsis fatalities. This aligns with Pillar I Excellent Science MSCA fellowship under the European Research Council and Key Strategic Orientation A with Cluster 1 of the Work Programme on technologies for healthy society.
During the reporting period, the project focused on the development and validation of a sensing and data-analysis framework for early detection of sepsis-related bacterial activity. A key technical activity was the design and fabrication of a custom gas pre-concentrator, which was successfully integrated with commercial gas sensors to enhance detection sensitivity. The integrated system was used to continuously monitor volatile gases released during bacterial growth in culture dishes relevant to sepsis diagnostics. These experiments demonstrated the feasibility of using low-cost, commercially available gas sensors for real-time monitoring of bacterial metabolism under standard clinical culture conditions.
In parallel, the project developed and applied advanced mathematical analysis methods to the recorded gas-sensor time-series data. Change-point detection algorithms were implemented to identify statistically significant changes in signal slope and amplitude and to associate these changes with bacterial growth dynamics. This analytical framework enabled the identification of empirical threshold values corresponding to bacterial concentrations above the clinically relevant infection threshold of approximately 10⁴ cfu/mL. In support of the planned in-vivo studies, a dedicated experimental protocol for mouse models of sepsis was prepared and successfully approved by the relevant bioethics committee.
The main scientific achievements of the project include the demonstration of an 18-fold increase in gas-sensor signal through the use of the integrated gas pre-concentrator, exceeding the performance targets defined in the project objectives. The combination of enhanced sensitivity and time-resolved analysis enabled the detection of bacterial infection within 2–10 hours, a substantial improvement over the conventional 24-hour culture-based diagnostic timeline, while fully maintaining the existing clinical protocol for bacterial cultivation. Furthermore, the project established robust empirical criteria linking signal changes detected by change-point analysis to clinically meaningful bacterial concentrations.
The outcomes of the project show that commercial gas sensors, when combined with a pre-concentrator and appropriate time-series analysis, can be effectively used for real-time monitoring of gas emissions from bacterial cultures. Change-point detection provides a powerful tool for early identification of infection-related metabolic activity, offering a rapid and early indicator that complements current clinical diagnostic workflows without requiring changes to established laboratory procedures.
In parallel, the project developed and applied advanced mathematical analysis methods to the recorded gas-sensor time-series data. Change-point detection algorithms were implemented to identify statistically significant changes in signal slope and amplitude and to associate these changes with bacterial growth dynamics. This analytical framework enabled the identification of empirical threshold values corresponding to bacterial concentrations above the clinically relevant infection threshold of approximately 10⁴ cfu/mL. In support of the planned in-vivo studies, a dedicated experimental protocol for mouse models of sepsis was prepared and successfully approved by the relevant bioethics committee.
The main scientific achievements of the project include the demonstration of an 18-fold increase in gas-sensor signal through the use of the integrated gas pre-concentrator, exceeding the performance targets defined in the project objectives. The combination of enhanced sensitivity and time-resolved analysis enabled the detection of bacterial infection within 2–10 hours, a substantial improvement over the conventional 24-hour culture-based diagnostic timeline, while fully maintaining the existing clinical protocol for bacterial cultivation. Furthermore, the project established robust empirical criteria linking signal changes detected by change-point analysis to clinically meaningful bacterial concentrations.
The outcomes of the project show that commercial gas sensors, when combined with a pre-concentrator and appropriate time-series analysis, can be effectively used for real-time monitoring of gas emissions from bacterial cultures. Change-point detection provides a powerful tool for early identification of infection-related metabolic activity, offering a rapid and early indicator that complements current clinical diagnostic workflows without requiring changes to established laboratory procedures.
Results
The project demonstrated that bacterial infections in culture can be detected significantly faster than current clinical practice while fully maintaining existing culture protocols. By combining low-cost commercial gas sensors, a custom gas pre-concentrator, and time-series change-point detection, the project achieved early identification of bacterial growth based on metabolic gas emissions rather than endpoint measurements. This represents a shift from static, delayed diagnostics toward continuous, real-time monitoring of bacterial cultures.
Indicative Impacts
The results of the project introduce a new early-indicator capability that can support clinicians in making timelier and more informed treatment decisions. Faster indication of bacterial growth has the potential to improve clinical outcomes by enabling earlier antibiotic selection, particularly in time-critical scenarios such as intensive care units, post-surgical monitoring, and severe infections.
Clinical microbiology laboratories and hospitals performing routine blood and urine cultures may benefit from reduced diagnostic timelines, improved laboratory throughput, and lower operational costs. From a patient perspective, earlier detection of infection may translate into reduced hospitalisation time, faster recovery, and improved quality of care for patients in both hospital and outpatient settings, including urinary tract infection cases managed at home.
Needs for Further Uptake and Success
To ensure further uptake and successful translation of the results, several steps are required. Protection of the underlying intellectual property at this early stage is essential to enable future commercialisation and engagement with industrial partners. Further research is needed using larger and more diverse bacterial populations to strengthen robustness, assess generalisability, and investigate whether bacterial species differentiation can be achieved using the same real-time data streams.
Additional validation in laboratory and relevant operational environments will be required to progress the technology beyond its current maturity level. Once validated, access to markets and financial support will be necessary to transition from a research prototype to a spin-off or industrial product. Although the proposed methodology does not alter existing clinical culture workflows, engagement with regulatory and standardisation bodies will be important to formally position the technology as an accepted early-indicator tool within clinical diagnostic practice.
Overview of Results
Overall, the project delivered a new diagnostic capability that enables real-time monitoring of bacterial cultures using low-cost sensing hardware and advanced time-series analytics. The results demonstrate that clinically relevant infection thresholds can be detected hours earlier than current standards without modifying established laboratory protocols. This positions the technology as a complementary early-warning tool with strong potential for clinical adoption, further research expansion, and future commercial exploitation.
The project demonstrated that bacterial infections in culture can be detected significantly faster than current clinical practice while fully maintaining existing culture protocols. By combining low-cost commercial gas sensors, a custom gas pre-concentrator, and time-series change-point detection, the project achieved early identification of bacterial growth based on metabolic gas emissions rather than endpoint measurements. This represents a shift from static, delayed diagnostics toward continuous, real-time monitoring of bacterial cultures.
Indicative Impacts
The results of the project introduce a new early-indicator capability that can support clinicians in making timelier and more informed treatment decisions. Faster indication of bacterial growth has the potential to improve clinical outcomes by enabling earlier antibiotic selection, particularly in time-critical scenarios such as intensive care units, post-surgical monitoring, and severe infections.
Clinical microbiology laboratories and hospitals performing routine blood and urine cultures may benefit from reduced diagnostic timelines, improved laboratory throughput, and lower operational costs. From a patient perspective, earlier detection of infection may translate into reduced hospitalisation time, faster recovery, and improved quality of care for patients in both hospital and outpatient settings, including urinary tract infection cases managed at home.
Needs for Further Uptake and Success
To ensure further uptake and successful translation of the results, several steps are required. Protection of the underlying intellectual property at this early stage is essential to enable future commercialisation and engagement with industrial partners. Further research is needed using larger and more diverse bacterial populations to strengthen robustness, assess generalisability, and investigate whether bacterial species differentiation can be achieved using the same real-time data streams.
Additional validation in laboratory and relevant operational environments will be required to progress the technology beyond its current maturity level. Once validated, access to markets and financial support will be necessary to transition from a research prototype to a spin-off or industrial product. Although the proposed methodology does not alter existing clinical culture workflows, engagement with regulatory and standardisation bodies will be important to formally position the technology as an accepted early-indicator tool within clinical diagnostic practice.
Overview of Results
Overall, the project delivered a new diagnostic capability that enables real-time monitoring of bacterial cultures using low-cost sensing hardware and advanced time-series analytics. The results demonstrate that clinically relevant infection thresholds can be detected hours earlier than current standards without modifying established laboratory protocols. This positions the technology as a complementary early-warning tool with strong potential for clinical adoption, further research expansion, and future commercial exploitation.