Abstract:This study presents the first implementation of multilayer neural networks on a memristor/CMOS integrated system on chip (SoC) to simultaneously detect multiple diseases. To overcome limitations in medical data, generative AI techniques are used to enhance the dataset, improving the classifier's robustness and diversity. The system achieves notable performance with low latency, high accuracy (91.82%), and energy efficiency, facilitated by end-to-end execution on a memristor-based SoC with ten 256x256 crossbar arrays and an integrated on-chip processor. This research showcases the transformative potential of memristive in-memory computing hardware in accelerating machine learning applications for medical diagnostics.

What problem does this paper attempt to address?

The main problems that this paper attempts to solve are the limitations of traditional blood - testing methods in disease diagnosis, including long processing time, limited accuracy, and low sensitivity. Specifically: 1. **Limitations of Traditional Blood Testing**: - Traditional methods require at least 15 minutes to separate serum, isolate target substances, and generate results. - Low sensitivity makes these methods usually only an auxiliary tool rather than a primary diagnostic means. 2. **The Need for Multi - Disease Detection**: - Most current disease - detection systems can only detect a single disease and are limited by hardware computing power. - Small datasets and high costs lead to over - fitting in deep neural networks and a decline in model reliability. 3. **Deficiencies in Existing Memristor - Based Systems**: - Most existing memristor - based systems are implemented in analog environments or on large experimental devices and cannot be integrated into portable devices, limiting practical applications. - These systems are mainly used for single - disease detection and fail to fully utilize the hardware potential for multi - disease detection. To solve the above problems, this research proposes a multi - disease detection system based on a memristor/CMOS integrated chip (SoC). The system achieves improvements in the following ways: - **Integrated Hardware and Software Design**: Utilize the nanofiber platform to enhance Raman signals and combine with the conditional diffusion model to generate high - quality data, improving the accuracy and robustness of the classifier. - **High Efficiency and Low Latency**: Through end - to - end execution, low latency, high accuracy (91.82%), and high energy efficiency are achieved. - **Simultaneous Detection of Multiple Diseases**: It can simultaneously detect multiple diseases such as acute myocardial infarction (AMI) and liver cancer. - **Data Augmentation Technique**: Use generative AI technology to expand the dataset, improving the diversity and generalization ability of the model. In conclusion, this research aims to develop a compact, efficient, and multi - functional disease - detection system to overcome the limitations of traditional methods and provide faster and more accurate solutions for medical diagnosis.

Multi-diseases detection with memristive system on chip

Hybrid Dual-Complementary Metal-Oxide-semiconductor/memristor Synapse-Based Neural Network with Its Applications in Image Super-Resolution

A Brain-Inspired In-Memory Computing System for Neuronal Communication Via Memristive Circuits.

Memristor-Based Hierarchical Attention Network for Multimodal Affective Computing in Mental Health Monitoring

Multichannel parallel processing of neural signals in memristor arrays

Efficient and self-adaptive in-situ learning in multilayer memristor neural networks

Neuronal Multi Unit Activity Processing with Metal Oxide Memristive Devices

Hardware Demonstration of SRDP Neuromorphic Computing with Online Unsupervised Learning Based on Memristor Synapses

Memristor-Based Artificial Chips

Memristor‐Based Analog Computation and Neural Network Classification with a Dot Product Engine

An efficient ANN SoC for detecting Alzheimer's disease based on recurrent computing

Memristive brain-like computing

Application of Mathematical Morphology Operation with Memristor-Based Computation-in-memory Architecture for Detecting Manufacturing Defects

Fully Hardware Memristive Neuromorphic Computing Enabled by the Integration of Trainable Dendritic Neurons and High‐Density RRAM Chip

Homogeneous Spiking Neuromorphic System for Real-World Pattern Recognition

In-Memory Computing With Memristor Arrays

A neuromorphic physiological signal processing system based on VO2 memristor for next-generation human-machine interface

Multimodal Neuromorphic Sensory-Processing System With Memristor Circuits for Smart Home Applications

Neurohybrid Memristive CMOS-Integrated Systems for Biosensors and Neuroprosthetics.

CMOS-integrated memristive non-volatile computing-in-memory for AI edge processors

Energy-efficient high-fidelity image reconstruction with memristor arrays for medical diagnosis