What does this research mean for the field?

Integrating a lightweight 1D CNN with a classic LMS adaptive filter achieves superior real-time noise suppression and diagnostic feature preservation in non-stationary biomedical signals compared to standalone LMS, RLS, and CNN-based methods. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

To develop an AI-augmented adaptive filtering framework that combines a lightweight 1D convolutional neural network with a conventional adaptive filter for real-time noise suppression in biomedical signals.

October 8, 2026National Journal of Signal and Image ProcessingOpen Access

AI-Augmented Adaptive Filtering for Real-Time Noise Suppression and Feature Enhancement in Biomedical Signals

Q: What are the key findings of this study?

The AI-augmented adaptive filtering framework achieved a higher Signal-to-Noise Ratio of 13.4 dB compared to 8.2 dB for the standalone LMS filter, demonstrating superior real-time noise suppression.

Q: What is the clinical evidence from this study?

Population: Biomedical signal noise (ECG and EEG) (n=109). Intervention: AI-augmented adaptive filtering framework (1D-CNN + LMS) vs. Standalone LMS, RLS, and CNN-only filters. Primary outcome: Signal-to-Noise Ratio (SNR) in dB.

View Full Paper

Ask AI

Bookmark

Key result

AI-augmented adaptive filtering boosts ECG signal-to-noise ratio by ~5 dB over standalone LMS.

n=109

Why the study?

Traditional adaptive filters like LMS perform well in stationary conditions but fail in non-stationary biomedical signal noise environments, limiting analysis accuracy.

Does an AI-augmented adaptive filtering framework improve noise suppression and feature preservation in biomedical signals compared to standalone conventional and deep learning filters?

Key Points

To develop an AI-augmented adaptive filtering framework that combines a lightweight 1D convolutional neural network with a conventional adaptive filter for real-time noise suppression in biomedical signals.
Integrated a lightweight 1D CNN noise prediction architecture with a standard Least Mean Squares (LMS) adaptive filter to dynamically modulate filter weights based on contextual noise profiles.
Evaluated performance on the MIT-BIH Arrhythmia and PhysioNet EEG Motor Movement benchmark datasets, with hardware implementation evaluated on an ARM Cortex-M4 processor.
Benchmarked the proposed hybrid model against standalone LMS, Recursive Least Squares (RLS), and CNN-only filtering models using SNR, RMSE, MAPE, and inference latency.
The hybrid LMS-CNN framework achieved an inference time of 3.1 ms (<4 ms) using quantized CNN inference on embedded hardware.
The hybrid method outperformed standalone LMS, RLS, and CNN-based filters across SNR, RMSE, and MAPE metrics while preserving essential clinical diagnostic features.

Structured PICO

Does an AI-augmented adaptive filtering framework improve noise suppression and feature preservation in biomedical signals compared to standalone conventional and deep learning filters?

Population

Evaluation of a signal processing framework using standard biomedical repositories, including 48 half-hour ambulatory ECG recordings and EEG recordings from 109 individuals.

Intervention

AI-enhanced adaptive filtering framework integrating a 1D Convolutional Neural Network (CNN) noise predictor with a classic Least Mean Squares (LMS) filter, quantized to 8-bit integers and run on an ARM Cortex-M4 embedded processor.

Comparator

Traditional standalone filtering models including Least Mean Squares (LMS), Recursive Least Squares (RLS), and a standalone CNN-based denoiser.

Outcome

Signal-to-Noise Ratio (SNR), Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and Inference Time.surrogate

Main Result

Absolute Event Rate: 13.4% vs 8.2%

A hybrid AI-augmented adaptive filter combining LMS and a lightweight 1D-CNN significantly improves real-time noise suppression in ECG and EEG signals while maintaining low latency suitable for edge-AI wearable devices.

Limitations

Currently limited to single-channel biomedical data analysis.
Further hardware-software co-design is needed to reduce power consumption for ultra-low-power applications.
Evaluated primarily on single-channel data, requiring future extension to multi-channel signals (e.g., EEG and EMG) using graph-based or multi-branch CNNs.
Lacks validation in clinical-grade real-time monitoring systems with patient-in-the-loop testing and cross-subject generalization analysis.

Implication

May enhance wearable ECG monitoring; leaves open clinical validation before practice adoption.

Abstract

Electrocardiograms (ECG), electroencephalograms (EEG), and electromyograms (EMG) are examples of biomedical signals that are frequently compromised by motion artifacts, the power-line interference, and physiological noise interferences, thus, limiting the accuracy of their analysis. Traditional adaptive filters that are in use such as the Least Mean Squares (LMS) filter work reasonably well in stationary situations but fail in non-stationary ones found in reality. The proposed paper introduces a new concept of AI-enhanced adaptive filtering framework which integrates conventional adaptive filtering algorithm with data-driven noise prediction model to improve real-time signal quality. The proposed technique incorporates the 1D CNN lightweight architecture with the classic LMS filter, making it possible to adjust filter weighting to contextual patterns of signals, in order to better detect and neutralize time-varying noise by the system. This architecture is justified using some benchmark biomedical datasets such as the MIT-BIH Arrhythmia and PhysioNet EEG Motor Movement set. Quantitative tests based on measures Signal-to-Noise Ratio (SNR), Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) demonstrate the high effectiveness of the suggested method in comparison with standalone LMS, RLS, and CNN-based filters. Notably, the design reaches real-time on embedded processors and would support wearable and edge-AI healthcare. The obtained results indicate that both noise suppression and preservation of diagnostic features improve significantly, providing a basis to further develop intelligent biomedical signal processing systems.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Soy et al. (2025) studied Biomedical signal noise (ECG and EEG) (n=109). AI-augmented adaptive filtering framework (1D-CNN + LMS) vs. Standalone LMS, RLS, and CNN-only filters was evaluated on Signal-to-Noise Ratio (SNR) in dB. The AI-augmented adaptive filtering framework achieved a higher Signal-to-Noise Ratio of 13.4 dB compared to 8.2 dB for the standalone LMS filter, demonstrating superior real-time noise suppression.

synapsesocial.com/papers/6ac71bf0ce9b52c5cbf931b2 https://doi.org/10.17051/njsip/01.04.01

Topics

Artificial intelligence in cardiology

View Full Paper

Ask AI

Bookmark

Implication

May enhance wearable ECG monitoring; leaves open clinical validation before practice adoption.

AI-Augmented Adaptive Filtering for Real-Time Noise Suppression and Feature Enhancement in Biomedical Signals

Key Points

Structured PICO

Main Result

Limitations

Implication

Abstract

Discussion

Cite This Study

Topics

AI-Augmented Adaptive Filtering for Real-Time Noise Suppression and Feature Enhancement in Biomedical Signals

Key Points

Structured PICO

Main Result

Limitations

Implication

Abstract

Discussion

Cite This Study

Keep reading

Topics

Keep reading