Reconstructing Lactate Biosensor Parameters for Sweat Analysis through Inverse Modelling
||Reconstructing Lactate Biosensor Parameters for Sweat Analysis through Inverse Modelling|
||NAISS Small Compute|
||Ivan Robayo <email@example.com>|
||Kungliga Tekniska högskolan|
||2023-03-20 – 2023-07-01|
Sweat analysis has gained increasing attention recently due to its potential for non-invasive monitoring of biomarkers in various health and performance-related applications. For example, lactate, a metabolite produced during exercise, has been identified as a critical biomarker for monitoring athletic performance. As a result, lactate biosensors have been developed to measure lactate levels in sweat, providing a promising tool for real-time lactate monitoring during exercise or other activities. Recently, our group has developed a lactate sensor using diffusion-limiting membranes, which avoids direct contact of the enzyme-based electrode with the sample solution. Thus, it enhances the linear range of response. However, the accuracy and reliability of lactate biosensors depend on various parameters such as the enzymatic reaction kinetics, mass transport properties, and sensor design, which can be difficult to determine experimentally.
Inverse modelling offers a computational approach to reconstructing these parameters based on measured sensor responses. Inverse modelling involves finding the parameters that best match the sensor response to the accurate lactate concentration in the sweat sample. This approach can optimise the sensor performance, improve the accuracy of the lactate measurements, and enable the detection of lactate levels at lower concentrations.
This project aims to investigate the factors that impact the performance of lactate biosensors, explicitly focusing on the enzymatic reaction kinetics and transport properties. We've developed a thorough phenomenological model considering 12 parameters, 11 species, and two phases (aqueous and membrane) to analyse the sensor's performance. To accurately characterise the model's performance, we plan to perform an inverse model of the amperometric responses. Our team will use R (4.1.2) and the FME, deSolve, and Reactran packages to develop and optimise the model. To avoid being trapped in a local minimum, we will use a Latin hypercube sampling method to guess initial parameter values. The Levenberg-Marquardt and BFGS algorithms will be used to optimise the parameters and use the global fitting procedure, which requires fitting responses under different conditions. It currently takes seven days on a desktop computer to perform this calculation on a single experimental response. We plan to leverage Tetralith's unique capabilities to reduce computing time and improve the project outcome, as global fitting requires more computational resources. Ultimately, this project will provide valuable insights into the design and optimisation of lactate biosensors, which have crucial applications in healthcare and sports performance monitoring.