Kruthika Kikkeri

Research Abstract:

Point-of-care (PoC) systems have shown to be instrumental for clinical medicine and disease control. However, while single-analyte qualitative PoC assays have shown substantial utility (COVID-19 rapid tests), quantitative multiplexed assays enable differential diagnosis, can monitor disease progression, and provide a more nuanced analysis of diseases. Yet, existing quantitative PoCs only cover a limited number of biomarkers that are abundant in blood, are expensive, or employ complex in-line steps. My research explores how we can utilize microfluidics, electronics, and machine learning to bridge this gap through the development of a tunable, inexpensive, sample-to-answer PoC system with clinically relevant analytical capabilities, rapid turnaround times, and user-friendliness. My initial work focused on the development of a sample-to-answer PoC workflow which incorporated each aspect of the testing pipeline through the integration of a blood sample acquisition device, on-chip blood-to-plasma separation, bead-based biomarker capture, microfluidic handling, and electronic readout into a 30-min assay. The workflow was validated by measuring blood samples spiked with clinically relevant concentrations of various cytokine biomarkers. To reduce costs and reliance on cleanroom fabrication, which often limits PoC translation, I then introduced the first prototyping approach that combines rapid (~mins), low-cost (<1USD) fabrication of tape-based valved microfluidics with multiplexed electrodes: Microfluidics via Inkjet-printing and Xurography (MINX). MINX integrates inkjet-printed carbon electrodes and tape-based microfluidic channels using widely accessible hobby machines, low-cost materials, and low-training thresholds. Using MINX, I fabricated a multiplexed PoC device for electrochemical biomarker detection. We envision that this technique can facilitate inexpensive, rapid prototyping capabilities and enable translation of PoC systems where microfluidic channels, valves, and/or electrodes are required. Finally, I am implementing intelligent controls for our PoC to enable real-time dynamic range and limit-of-detection modulation. Due to the heterogeneity of patient profiles and their responses to disease, target ranges for detection can span 7 orders-of-magnitude (OM). However, assay development to expand these systems beyond 3-4 OMs is non-trivial and often laborious. Instead, I am applying a reinforcement learning model for automated real-time adjustment of operational parameters to dynamically adjust to inter- and intra-patient variations, as well as new analytes. PoCs are and will continue to be ubiquitous in healthcare. However, for PoCs to remain paced with current and emerging diseases, we must tackle the challenges of adaptability, robustness, scalability, and accessibility. Moreover, while most of this work centers on PoCs, each technique can be broadly applied for general microfluidic challenges.

Bio:

Kruthika Kikkeri is a PhD candidate in Electrical Engineering at MIT with an expected graduation date of May 2024. She is advised by Professor Joel Voldman in the Biological Microtechnology and BioMEMS Group under MIT’s Research Laboratory for Electronics (RLE) and Microsystems Technology Laboratories (MTL). Kruthika’s research has focused on the integrating principles from microfluidics, electronics, and machine learning to develop low-cost, adaptive point-of-care (PoC) platforms for biomarker detection. She has been the recipient of the NSF Graduate Research Fellowship, MathWorks Fellowship, Takeda Fellowship, Whitaker Health Sciences Fellowship, MIT Fett Fellowship, and Virginia Tech William Webber Fellowship. Previously, Kruthika received her BS and MS from Virginia Tech in Electrical Engineering and interned at Lockheed Martin, Boeing, and GE Aviation.