Print Email Facebook Twitter Towards Continuous Monitoring of Volatile Organic Compounds in the Gastrointestinal Tract Title Towards Continuous Monitoring of Volatile Organic Compounds in the Gastrointestinal Tract Author de Gruijl, David (TU Delft Electrical Engineering, Mathematics and Computer Science) Contributor Langendoen, K.G. (mentor) Traverso, Giovanni (graduation committee) Degree granting institution Delft University of Technology Programme Electrical Engineering | Embedded Systems Date 2023-08-29 Abstract Much like wearable devices today, ingestible devices have emerged as a promising platform for continuous health monitoring, and potentially even intervention. Recent research has demonstrated the feasibility of ingestible devices with a retention mechanism, enabling them to remain in the stomach for weeks. Equipping these devices with sensors capable of measuring complex biomarkers, would open up an entirely new era of continuous health monitoring.This research focuses on the design of a gas sensor array for a retentive ingestible device, targeting the measurement of volatile organic compounds in the gastrointestinal tract. The research specifically addresses the outpatient treatment of alcoholism, in collaboration with the emergency medicine department at Brigham and Women's Hospital, Harvard Medical School. Current treatment relies on manual registration of drinking behavior, and the incorporation of an ingestible device that is able to continuously monitor drinking behavior enables more accurate behavioral assessment, and more targeted support and treatment.Given the novelty of this approach, a sensor array capable of assessing ethanol concentrations in the air was gradually subjected to more complex tasks. Furthermore, a key aspect of the research was the design of an artificial gastric environment, that replicates the conditions and challenges that the sensor array would encounter within the human body. This experimental environment played a crucial role by providing a realistic testing platform without the need for a fully functional ingestible device, which can be incredibly challenging and resource-intensive to manufacture.To ensure functionality in humid conditions, the sensors are encapsulated using parylene C and polycaprolactone (PCL) - biocompatible materials commonly used in the design of biomedical devices. The impact of these encapsulation methods on the sensors is thoroughly assessed, to determine their viability for in-vivo applications.The findings of this study reveal that employing a convolutional neural network can enable the accurate measurement of ethanol in air, using off-the-shelf air quality sensors and algorithms with low computational complexity. It is worth highlighting that the neural network is capable of performing inference directly on the ingestible device. Furthermore, initial results show that a combination of parylene C and PCL, achieved through dip-coating in a PCL-dichloromethane solvent, yields a sensor capable of reliably distinguishing ethanol percentages from 4 to 11 volume percentages, while continually submerged in a self-designed artificial gastric environment.Overall, this research contributes to the advancement of ingestible sensors and their potential for continuous health monitoring, with a use case in alcoholism treatment. The outcomes show the potential to produce a sensor array encapsulated in biocompatible materials, with a data-driven sensor fusion algorithm that is deployable on-device, which brings us closer to practical in-vivo applications. Additionally, the design and utilization of an artificial gastric environment establishes a solid foundation for future studies and the data generation that is vital for this technology. Subject Biomedical DevicesTinyMLBiomarkerVolatile Organic Compoundselectronic noseEmbedded SystemsParylene CPCLgas sensinggas sensoralcoholismalcoholEdge AISmart PillSensor array To reference this document use: http://resolver.tudelft.nl/uuid:ba9a28ef-7146-4b0e-85f0-4f7cd1ef343e Part of collection Student theses Document type master thesis Rights © 2023 David de Gruijl Files PDF deGruijl_David_MSc_Thesis_24.pdf 12.89 MB Close viewer /islandora/object/uuid:ba9a28ef-7146-4b0e-85f0-4f7cd1ef343e/datastream/OBJ/view