Minimally-invasive technology for personalized nutritional monitoring

Project Summary/Abstract Monitoring and optimizing dietary intake are important for precision nutrition in the treatment of clinical conditions (diabetes, obesity, kidney/liver failure, etc) as well as attenuating loss of muscle function and mass during aging. However, current methods for personal health monitoring do not provide reliable measurement of

macronutrient intake or availability of metabolites after digestion. Thus, this project aims to address this gap by developing multi-analyte sensing devices based on an innovative combination of insertable optical reporters, wearable readers, and advanced computational methods. These devices will provide continuous/on-demand

measurement of metabolites in interstitial fluid (ISF), then use those measures to predict macronutrient intake. The project is organized into three specific aims: Aim 1 will involve collecting blood samples and ISF samples (via microdialysis) from healthy human subjects consuming fresh meals and then using machine learning

methods to establish a quantitative relationship between macronutrient intake and the resulting blood and ISF levels of metabolites. Fresh meals will be designed by the investigators to have three different levels of protein and carbohydrates. Data will be analyzed to establish quantitative relationships between compartmental

concentrations, identifying proportionality coefficients and temporal lag/lead parameters, and develop machine- learning models to predict macronutrients in meals from the plasma and ISF concentrations of metabolites. Aim 2 will proceed in parallel with Aim 1, focusing on modifying current oxygen and glucose-sensing devices to

measure amino acids. Extensive benchtop testing will be used to verify accuracy and precision of calibration based on fusion of sensor data streams. Aim 3 will then focus on deploying and testing a removable version of the multi-analyte sensing system in human subjects. Participants will be fitted with the sensors, a microdialysis

catheter, and a continuous glucose monitor. Subjects will consume fresh meals as well as macronutrient- matched frozen meals, matched for contents, to verify that the system (sensors and algorithms) also work for meals that are more representative of the diet for many Americans, especially in low-income communities. The

computational algorithms developed in Aim 1 will be used to predict macronutrient availability from sensor data as well as microdialysis. Sex, skin color, and health indicators (BMI, HbA1c, etc) will be factored into analyses to assess whether they substantially affect performance of instrumentation and algorithms.

This project will bring together a team of biomedical engineers, computer scientists, and nutrition & metabolism researchers to develop instrumentation and collect data from human subjects that will yield: new knowledge and insight on the relationship between nutritional intake and systemic metabolite levels (Aim 1); advances in

technology for sensing metabolites as nutritional biomarkers (Aim 2); evaluation of the sensing technology and computational models in diverse human subjects consuming common meals. (Aim 3). Success in this project will advance research on metabolism and support development of new approaches to personalized nutrition.