Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for NeuroBridge: AI-Powered Biomedical Intelligence

🧠 NeuroBridge: AI-Powered Biomedical Intelligence

The Problem

Biomedical signals contain valuable information about human movement and physiological activity, but raw sensor data is difficult for non-specialists to understand. Conventional processing can also require significant computation outside the sensing device, creating opportunities for higher latency and reduced edge intelligence.

We wanted to explore a different approach: what if biomedical signals could be processed intelligently at the edge using dedicated hardware, and the resulting information could then be presented to people in a simple, understandable way?

Our Solution

NeuroBridge is an FPGA-based neuromorphic biomedical intelligence platform designed to process biomedical signals at the edge and transform them into meaningful, human-readable health information.

The core of NeuroBridge is a hardware processing pipeline implemented on an FPGA. Instead of sending raw biomedical data directly to a conventional software system, the FPGA performs signal processing, feature extraction, and neuromorphic inference before producing structured results.

Our planned end-to-end system is:

Biomedical Sensors → Signal Acquisition → FPGA Processing → Neuromorphic Analysis → Data Analytics → Health Report → AI Assistant

The FPGA acts as the computational core, while the application layer provides visualization, reporting, and interaction.

What NeuroBridge Processes

The platform is designed around biomedical and movement-related signals such as:

• EMG signals for muscle activity analysis
• IMU data for movement and posture analysis
• Extracted signal features
• Neuromorphic model outputs
• Classification results and confidence values
• Processing and hardware-performance metrics

The processed information can be converted into visual analytics such as signal waveforms, feature measurements, trends, and model outputs.

Neuromorphic Processing

A key aspect of NeuroBridge is its use of neuromorphic computing concepts.

The signal-processing pipeline extracts meaningful features from biomedical signals and passes them into a spiking/neuromorphic processing stage for pattern classification.

This allows us to investigate how event-driven computation can be used for low-latency biomedical signal processing on resource-constrained edge hardware.

From FPGA Data to a Human-Readable Report

One of our main goals is to bridge the gap between engineering data and understandable health information.

The NeuroBridge application receives processed data from the FPGA and organizes it into a structured session.

The application can then generate a report containing:

• Patient/session information
• Signal measurements
• Extracted features
• Neuromorphic classification
• Confidence values
• Historical trends
• FPGA processing metrics
• Plain-language explanations

This transforms a collection of numerical hardware outputs into a report that is much easier to understand.

NeuroBridge AI

We are also developing an AI interaction layer that allows users to ask questions about their generated reports.

For example:

"What does this EMG measurement mean?"

"Explain my report in simple language."

"What is median frequency?"

"Why did the classification change between sessions?"

The AI is intended to act as an educational and report-explanation assistant, helping users understand biomedical terminology and the information contained in their reports.

The system is not intended to replace a qualified medical professional or provide an independent medical diagnosis.

How We Built It

The hardware processing core was developed using FPGA-based digital design and RTL development.

The overall architecture combines:

• FPGA-based signal processing
• Digital filtering
• Feature extraction
• Neuromorphic/spiking computation
• Biomedical signal analysis
• Embedded communication
• Data visualization
• Automated report generation
• AI-assisted interaction

The FPGA prototype provides the hardware foundation, while the application layer is being developed to communicate with the processing core and present its results.

Why FPGA?

FPGA implementation allows us to prototype and evaluate the architecture before considering a dedicated silicon implementation.

It gives us the ability to measure important hardware characteristics such as:

• Processing latency
• Throughput
• Resource utilization
• Memory usage
• DSP utilization
• Estimated power

This also provides a pathway toward future ASIC implementation and optimization.

Challenges

One of our biggest challenges is connecting multiple domains into one system.

Biomedical signals are noisy and continuous, while digital hardware requires carefully designed data paths and timing. At the same time, the results produced by the FPGA must be translated into information that a normal user can understand.

Another challenge is balancing:

accuracy + latency + hardware resources + usability

rather than optimizing only one of these factors.

What We Learned

Through NeuroBridge, we are exploring the complete path from a biomedical signal to an intelligent edge-computing system.

We learned that building a useful healthcare technology prototype is not only about creating an AI model. It requires integration across sensing, signal processing, hardware acceleration, neuromorphic computation, software, visualization, and responsible communication of results.

Future Vision

Our long-term goal is to evolve NeuroBridge into a low-power edge biomedical intelligence platform.

The current FPGA implementation serves as a prototype for an architecture that could eventually be optimized for ASIC implementation, enabling more compact and potentially energy-efficient biomedical processing hardware.

Our vision is:

Sense → Process → Understand → Explain

all while keeping intelligent computation as close to the source of the biomedical data as possible.

NeuroBridge connects biomedical sensing with intelligent edge hardware and human understanding.

Built With

• artificial
• biomedical
• computing
• digital
• edge
• embedded
• esp32
• fpga
• healthcare
• healthtech
• intelligence
• iot
• machine
• matlab
• networks
• neural
• neuromorphic
• processing
• python
• signal
• spiking
• verilog
• vivado
• vlsi
• xilinx