Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Signal Quality Auditing for Time-series Data

Chufan Gao, Nicholas Gisolfi, Artur Dubrawski

TL;DR

This paper addresses time-series data quality in AI-driven Predictive Maintenance by profiling a broad set of signal quality indices (SQIs) and evaluating denoising strategies. It delivers an open-source toolkit implementing SQIs for ECG, Pleth, and other vital signs, and validates them on PhysioNet 2011 ECG data and other benchmarks. Key findings include that a combined SQI feature set outperforms individual SQIs across classification and outlier-detection tasks, that Isolation Forest and Autoencoders leverage SQIs effectively for anomaly detection, and that CNN-based denoising improves real-vs-artifact alert discrimination. The work provides reproducible baselines and a generalizable framework for QA of time-series measurements in complex systems.

Abstract

Signal quality assessment (SQA) is required for monitoring the reliability of data acquisition systems, especially in AI-driven Predictive Maintenance (PMx) application contexts. SQA is vital for addressing "silent failures" of data acquisition hardware and software, which when unnoticed, misinform the users of data, creating the risk for incorrect decisions with unintended or even catastrophic consequences. We have developed an open-source software implementation of signal quality indices (SQIs) for the analysis of time-series data. We codify a range of SQIs, demonstrate them using established benchmark data, and show that they can be effective for signal quality assessment. We also study alternative approaches to denoising time-series data in an attempt to improve the quality of the already degraded signal, and evaluate them empirically on relevant real-world data. To our knowledge, our software toolkit is the first to provide an open source implementation of a broad range of signal quality assessment and improvement techniques validated on publicly available benchmark data for ease of reproducibility. The generality of our framework can be easily extended to assessing reliability of arbitrary time-series measurements in complex systems, especially when morphological patterns of the waveform shapes and signal periodicity are of key interest in downstream analyses.

Signal Quality Auditing for Time-series Data

TL;DR

This paper addresses time-series data quality in AI-driven Predictive Maintenance by profiling a broad set of signal quality indices (SQIs) and evaluating denoising strategies. It delivers an open-source toolkit implementing SQIs for ECG, Pleth, and other vital signs, and validates them on PhysioNet 2011 ECG data and other benchmarks. Key findings include that a combined SQI feature set outperforms individual SQIs across classification and outlier-detection tasks, that Isolation Forest and Autoencoders leverage SQIs effectively for anomaly detection, and that CNN-based denoising improves real-vs-artifact alert discrimination. The work provides reproducible baselines and a generalizable framework for QA of time-series measurements in complex systems.

Abstract

Signal quality assessment (SQA) is required for monitoring the reliability of data acquisition systems, especially in AI-driven Predictive Maintenance (PMx) application contexts. SQA is vital for addressing "silent failures" of data acquisition hardware and software, which when unnoticed, misinform the users of data, creating the risk for incorrect decisions with unintended or even catastrophic consequences. We have developed an open-source software implementation of signal quality indices (SQIs) for the analysis of time-series data. We codify a range of SQIs, demonstrate them using established benchmark data, and show that they can be effective for signal quality assessment. We also study alternative approaches to denoising time-series data in an attempt to improve the quality of the already degraded signal, and evaluate them empirically on relevant real-world data. To our knowledge, our software toolkit is the first to provide an open source implementation of a broad range of signal quality assessment and improvement techniques validated on publicly available benchmark data for ease of reproducibility. The generality of our framework can be easily extended to assessing reliability of arbitrary time-series measurements in complex systems, especially when morphological patterns of the waveform shapes and signal periodicity are of key interest in downstream analyses.
Paper Structure (7 sections, 2 figures, 5 tables)

This paper contains 7 sections, 2 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Methodology and Experiments
  3. Q1: Which SQI methods yield top-performing models in signal quality classification?
  4. Q2: Which model classes benefit most from SQIs in an outlier detection task?
  5. Q3: How useful are SQIs at denoising time-series?
  6. Q4: What is the effect of ECG Denoising on Real vs Artifactual Alert Detection?
  7. Discussion and Conclusion

Figures (2)

  • Figure 1: ECG Denoising Results example from the CNN autoencoder (the best model in terms of Mean Squared Error). The blue is the original signal, and the orange is the cleaned signal.
  • Figure 2: Plots of Receiver Operating Characteristic (ROC) curves (with AUC and stanrdard deviation) for Heart Rate alerts, with and without additional SQIs calculated on raw (represented by the blue line) and denoised ECG (represented by the orange line).