• Corpus ID: 3900713

beatDB : A Large Scale Waveform Feature Repository

@inproceedings{Dernoncourt2013beatDBA,
  title={beatDB : A Large Scale Waveform Feature Repository},
  author={Franck Dernoncourt and Kalyan Veeramachaneni},
  year={2013}
}
For typical physiological waveform studies, researchers define a study group within which they designate case and controls. They extract the group’s waveforms, filter the signals, pre process them and extract features before iteratively executing, evaluating and interpreting a pre-selected machine learning algorithm with metrics such as area under the curve and analyses such as variable sensitivity. 

Figures and Tables from this paper

BeatDB v3 : a framework for the creation of predictive datasets from physiological signals

The goal of this project is primarily to modify BeatDB to support multi-channel waveform data like EEG and accelerometer data and to make the project more flexible to modification by researchers.

Gaussian Process-Based Feature Selection for Wavelet Parameters: Predicting Acute Hypotensive Episodes from Physiological Signals

A scalable, robust system to find the best wavelet parameters using Gaussian processes (GPs) is presented and it is demonstrated that the use of GPs reduces the selection effort by a factor of 3 compared with a naive grid search.

PhysioMiner: A Scalable Cloud Based Framework for Physiological Waveform Mining

PhysioMiner is tested on a large dataset of electrocardiography (ECG) from 6000 patients in the MIMIC database and an area under the curve of 0.693 was achieved for an acute hypotensive event prediction from the ECG waveform alone.

Large-Scale Methodological Comparison of Acute Hypotensive Episode Forecasting Using MIMIC2 Physiological Waveforms

From the dataset extracted from the large ICU physiological waveform repository of MIMIC2 database, it is found that both Bayesian network and k-nearest neighbor-based predictors are effective for predicting AHE and their performances improve with increasing training dataset size.

Physiological time series retrieval and prediction with locality-sensitive hashing

This thesis develops highly efficient similarity-based methods that make it practical to search massive physiological time series repositories to rapidly identify waveforms similar to those from a given individual, and empirically demonstrates that LSH based retrieval and prediction methods vastly speed up querying time while sacrificing only a trivial amount of accuracy as a cost.

Sequential short-text classification with neural networks

This thesis introduces several algorithms to perform sequential short-text classification, which outperform state-of-the-art algorithms and proposes several natural language processing methods based on artificial neural networks to facilitate the completion of systematic reviews.

Teaching a computer to be a data scientist

Gigabeats, which judiciously pre-computes features per beat for reuse across investigations, and offers parameterizations that allow for task-specific computing and storage tradeoff decisions, resolves questions such as: what data abstractions make for efficient data storage, retrieval and processing.

References

SHOWING 1-2 OF 2 REFERENCES

Prediction of acute hypotensive episodes using neural network multi-models

  • J. HenriquesT. Rocha
  • Computer Science
    2009 36th Annual Computers in Cardiology Conference (CinC)
  • 2009
This work proposes the application of generalized regression neural network multi-models to the prediction of acute hypotensive episodes (AHE) occurring in intensive care units, and demonstrates the effectiveness of this strategy in the context of PhysioNet-Computers in Cardiology Challenge 2009.

A signal abnormality index for arterial blood pressure waveforms

A signal abnormality index (SAI) algorithm is presented that detects abnormal beats in ABP waveforms by intelligently setting constraints on physiologic, noise/artifact, and beat-to-beat variation values.