Heart age estimated using explainable advanced electrocardiography

Lindow, Thomas; Palencia-Lamela, Israel; Schlegel, Todd T.

doi:10.1101/2021.12.08.21267378

Cited by 1 publication

(3 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…If artifacts are uncorrelated with the outcome of interest, one might choose to leave them unprocessed and thereby potentially even increase the robustness of the learned representation. However, prior studies have shown that peripheral signals and EEG artifacts can be systematically modulated in different patient groups (Golding et al, 2006;Jongkees & Colzato, 2016;Lindow et al, 2023;Wilkinson & Nelson, 2021). This motivated us to systematically study the relationship between different EEG components attributed to CNS versus peripheral generators, which typically differ in terms of spectral and spatial patterns.…”

Section: /54mentioning

confidence: 99%

“…For at least three reasons, we hold that regardless of ICA quality, one has to consider that artifacts are predictive and risk diluting CNS biomarkers: 1) prior knowledge of the change in peripheral and body variables in aging and pathology (Golding et al, 2006;Jongkees & Colzato, 2016;Lage et al, 2020;Lindow et al, 2023;Wilkinson & Nelson, 2021). 2) The sensitivity of machine learning models to pick up even weak and hidden patterns.…”

Section: Artifact Removal Is Essential For Learning Interpretable Bio...mentioning

confidence: 99%

“…eye blinks, muscle & cardiac activity) that are well known to leak into the EEG and can overshadow the brain signal of interest. This is particularly problematic as these peripheral sources are often modified by medical conditions and other individual factors and might therefore be predictive too (Golding et al, 3/54 2006;Lage et al, 2020;Lindow et al, 2023), which is entirely ignored if EEG is left unprocessed as was recently advocated for (Delorme, 2023). Depending on the relationship between artifacts and variables of interest, high-capacity ML techniques such as deep neural networks may automatically learn how to filter out artifact-generating sources as irrelevant noise, or, instead will use non-brain information to minimize prediction error (Jochmann et al, 2023).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Machine learning of brain-specific biomarkers from EEG

Bomatter,

Paillard,

Garces

et al. 2023

Preprint

View full text Add to dashboard Cite

Electroencephalography (EEG) has a long history as a clinical tool to study brain function, and its potential to derive biomarkers for various applications is far from exhausted. Machine learning (ML) can guide future innovation by harnessing the wealth of complex EEG signals to isolate relevant brain activity. Yet, ML studies in EEG tend to ignore physiological artifacts, which may cause problems for deriving biomarkers specific to the central nervous system (CNS). We present a framework for conceptualizing machine learning from CNS versus peripheral signals measured with EEG. A common signal representation across the frequency spectrum based on Morlet wavelets allowed us to define traditional brain activity features (e.g. log power) and alternative inputs used by state-of-the-art ML approaches (covariance matrices). Using more than 2600 EEG recordings from large public databases (TUAB, TDBRAIN), we studied the impact of peripheral signals and artifact removal techniques on ML models in exemplary age and sex prediction analyses. Across benchmarks, basic artifact rejection improved model performance whereas further removal of peripheral signals using ICA decreased performance. Our analyses revealed that peripheral signals enable age and sex prediction. However, they explained only a fraction of the performance provided by brain signals. We show that brain signals and body signals, both reflected in the EEG, allow for prediction of personal characteristics. While these results may depend on specific prediction problems, our work suggests that great care is needed to separate these signals when the goal is to develop CNS-specific biomarkers using ML.

show abstract

Section: /54mentioning

confidence: 99%