1. Impedance Sensing Technology

Measurement technology centered on electrochemical impedance spectroscopy (EIS). It extracts complex impedance behavior from frequency response characteristics to non-destructively analyze dynamics associated with physical and chemical changes in the target as electrical signals.

2. Modeling & Simulation Technology

Mathematical modeling and numerical simulation technology of various physical phenomena. We express and understand phenomena through mathematical formulations, validating, visualizing, and predicting them via multi-physics simulations.

3. Analog Electronic Circuit Technology

Hardware configuration technology based on CMOS integrated circuit design. In addition to theoretical circuit design, we promote design automation assisted by machine learning algorithms.

4. AI, Machine Learning & Quantum Computing

Accelerating processing and optimization in data analysis and hardware design through the application of machine learning and quantum computing.

Computational Research on Non-Invasive Bio-Sensing via EIS and AI

Our research leverages Electrochemical Impedance Spectroscopy (EIS) for non-invasive, label-free cellular monitoring. We specialize in a hybrid approach combining numerical simulation and machine learning algorithms.

Two Core Technologies:

Advanced Numerical Simulation, Data Analysis, and Prediction

• Exploring Physical Phenomena: Modeling complex impedance at the bio-interface to theoretically explore cellular dynamics, such as adhesion and antigen-antibody reactions.
• Analysis & Prediction: Extracting physically meaningful features from datasets and predicting experimental outcomes based on high-fidelity models.
• Optimal Sensor Design: Theoretically deriving high-sensitivity electrode architectures through parametric simulation.

Advanced Data Analytics via Machine Learning

• Intelligent Data Analysis: Applying machine learning to multi-dimensional EIS datasets to extract subtle physiological features beyond conventional analysis methods.

Collaborative Framework

While we lead the theory and analytics, we maintain strong partnerships with experimental labs for cell culture and validation. This synergy between computation and experiment accelerates breakthroughs in drug discovery and medicine.

Relevant Publications

• Yusuke Sugahara and Shigeyasu Uno, "An analytic equation for single cell electrochemical impedance spectroscopy with a dependence on cell position", AIP Advances , vol.13, issue 9, p. 095315 (2023). (https://doi.org/10.1063/5.0166409)
• Masataka Shiozawa and Shigeyasu Uno, "An analytical formula for determining the electrical impedance between a single adherent cell and sensor substrate", Jpn. J. Appl. Phys. vol. 61, p. 117001 (2022). (https://doi.org/10.35848/1347-4065/ac9877)
• Shigeyasu Uno, "Electrochemical Impedance Sensor for Non-invasive Living Cell Monitoring toward CMOS Cell Culture Monitoring Platform", 4th International Symposium on Devices, Circuits and Systems (ISDCS2021), March 3, 2021 (Hiroshima: virtual), Session 3, no1. (10.1109/ISDCS52006.2021.9397893)
• Yuhki Yanase, Kyohei Yoshizaki, Kaiken Kimura, Tomoko Kawaguchi, Michihiro Hide and Shigeyasu Uno, "Development of SPR Imaging Impedance Sensor for Multi-Parametric Living Cell Analysis ", Sensors, vol. 19, issue 9, p. 2067 (2019). (10.3390/s19092067)
• Shigeyasu Uno, "Modeling and Simulation of Electrochemical Biosensors based on CMOS LSI Chips", IEEE 3rd Electron Devices Technology and Manufacturing (EDTM2019), March 15, 2019 (Singapore), Model 4, 3. (10.1109/EDTM.2019.8731254)
• Shigeyasu Uno, "Electrochemical Biosensors based on CMOS LIS Chips", 233rd ECS Meeting, May 15, 2018 (Seattle, USA), H02-1471.

Research on CMOS Integrated Circuit Design Automation via Machine Learning

Our research focuses on automating and accelerating CMOS integrated circuit design—historically dependent on expert intuition—using advanced AI. We specialize in an approach that integrates circuit theory with machine learning algorithms to redefine the design flow.

Core Technology: Advanced Data Analytics and Design Automation via Machine Learning

• Intelligent Analysis & Optimization: Applying machine learning to analyze vast simulation datasets. We explore complex parameter spaces to logically derive optimal device sizes and bias conditions that satisfy multi-objective specifications.
• Automated Topology Generation: Utilizing generative models and graph representation learning to automatically generate and optimize circuit topologies, enabling the discovery of innovative architectures beyond conventional design methods.

Future Outlook & Collaboration

While our primary focus is on theory, algorithms, and software development, we plan to expand our scope to include chip fabrication and measurement for empirical validation. Through collaborations with academic and industrial partners, we aim to create a revolutionary design platform that contributes to the advancement of next-generation semiconductor and system development.

Application of Sparse Sensor Placement Optimization to Diverse Sensing Challenges

This project focuses on Sparse Sensor Placement (SSPO) optimization to identify the most informative elements within large-scale candidate spaces. Based on the concept of "uncovering low-dimensional representations of phenomena," we are also expanding our scope to non-linear models such as Variational Autoencoders (VAEs), which are machine learning models using neural networks.

Core Technology: Dimensionality Reduction and Measurement Efficiency

• Extracting Low-Dimensional Structures: Developing algorithms to capture the essence of physical systems, enabling the reconstruction of the entire field from a minimal number of sensors.
• Application to Experimental Datasets: Applying SSPO to various real-world datasets to accurately predict or reconstruct unobserved states from sparse observations.
• Versatile Applications: Beyond spatial placement, our methods are applied to frequency selection in impedance spectroscopy and time-series analysis for multi-sensor arrays.

Specific Application Examples:

• Operational Cost Reduction: Driving only a subset of optimal sensors to predict data for idle sensors, making it possible to drastically reduce power consumption and costs in continuous monitoring for factories, buildings, and outdoor environments.
• Resource Optimization: Using a few mobile sensors to scan and identify optimal locations for localized monitoring, aimed at efficient broad-area surveillance with limited drones.
• Measurement Time Reduction: Selecting optimal frequencies to reconstruct full-spectrum data, aiming to replace expensive, slow, high-precision measurements with affordable, high-speed alternatives.

Collaborative Framework

Our core focus is on the implementation and improvement of mathematical methods and algorithms for real-world problems. For actual measurement data and field validation, we maintain strong partnerships with experimental labs. We aim to overcome the limitations of conventional sensing through the synergy of physical insight and advanced analytics.

Relevant Publications

• Shigeyasu Uno, ”Reconstruction of Full Time Series Data from Gas Sensor Array by Partial Measurement Using Sparse Sensor Placement Optimization Method”, IEEE Sensors 2025, October 20, 2025 (Vancouver), 5659. (10.1109/SENSORS59705.2025.11331262)

Optimization of Gate-Based Quantum Circuit Design

This project focuses on the automated optimization of quantum circuit design, which is essential for implementing quantum algorithms on gate-based quantum computers. We specialize in an approach that merges advanced deep learning models with quantum information theory to maximize the utilization of limited quantum resources.

Core Technology: Synergy of Quantum Information Theory and Machine Learning

• Circuit and Algorithm Optimization: Utilizing machine learning models such as Graph Neural Networks (GNNs) and Transformers to implement combinatorial optimization problems, quantum gate placement and connectivity are optimized for best performance. We develop methods to minimize computational errors and maximize the efficiency of quantum algorithms.
• Simulation and Model Validation: Conducting hybrid validation by combining quantum behavior simulations with various machine learning architectures. We derive optimal learning models based on theoretical performance predictions.
• Direct Hardware Verification: Personally operating and evaluating optimized circuits on various real-world quantum processors available via cloud platforms to conduct direct performance assessments.

Future Outlook

Our core focus is on the implementation and improvement of mathematical methods and algorithms for real-world problems. Looking ahead, we aim to implement these solutions to solve real-world challenges, such as combinatorial optimization problems. We strive to create a revolutionary software infrastructure that accelerates the practical application of quantum computing through the synergy of computational science and AI.