April 18, 2019Open Access

Explaining Deep Classification of Time-Series Data with Learned Prototypes

Q: What is the clinical evidence from this study?

Study design: Other. Population: Bradycardia and Apnea of prematurity (n=9458). Intervention: Deep classification model with prototype diversity penalty vs. Baseline model without prototype diversity penalty. Primary outcome: Classification accuracy for ECG bradycardia.

Resultado clave

A deep classification model with a prototype diversity penalty achieved a classification accuracy of 93.1% for detecting bradycardia in ECG waveforms, compared to 92.1% with the baseline model.

PICO estructurado

Población

Two-dimensional time-series data including electrocardiogram (ECG) waveforms from preterm infants, respiration waveforms, and audio waveforms

Intervención

Deep learning models with learned prototypes optimized for increased prototype diversity and robustness

Comparador

Existing deep learning models

Resultado

Classification decisions and explainable insights (detection of clinical bradycardia, apnea of prematurity, and spoken digits)

Leveraging learned prototypes in deep learning models can provide explainable insights for time-series classification tasks, such as detecting bradycardia in preterm infant ECGs.

Resultado numérico

Tasa de eventos absoluta: 93.1% vs 92.1%

Limitaciones

High number of loss terms creates a trade-off between prototype interpretability and model accuracy.
A small number of prototypes achieves near-perfect reconstruction at the cost of classification accuracy, while a large number achieves higher accuracy but yields noisy prototypes.

Resumen

The emergence of deep learning networks raises a need for explainable AI so users and domain experts can be confident applying them to high-risk. In this paper, we leverage data from the latent space induced by learning models to learn stereotypical representations or "prototypes" training to elucidate the algorithmic decision-making process. We study leveraging prototypes effect classification decisions of two dimensional-series data in a few different settings: (1) electrocardiogram (ECG) to detect clinical bradycardia, a slowing of heart rate, in preterm, (2) respiration waveforms to detect apnea of prematurity, and (3) waveforms to classify spoken digits. We improve upon existing models by for increased prototype diversity and robustness, visualize how prototypes in the latent space are used by the model to distinguish, and show that prototypes are capable of learning features on two time-series data to produce explainable insights during tasks. We show that the prototypes are capable of learning-world features - bradycardia in ECG, apnea in respiration, and in speech - as well as features within sub-classes. Our novel work learned prototypical framework on two dimensional time-series data to explainable insights during classification tasks.

Me gusta

Guardar

Ver artículo completo

Me gusta

Guardar

Ver artículo completo

Cite This Study

Gee et al. (Thu,) conducted a other in Bradycardia and Apnea of prematurity (n=9,458). Deep classification model with prototype diversity penalty vs. Baseline model without prototype diversity penalty was evaluated on Classification accuracy for ECG bradycardia. A deep classification model with a prototype diversity penalty achieved a classification accuracy of 93.1% for detecting bradycardia in ECG waveforms, compared to 92.1% with the baseline model.

synapsesocial.com/papers/6a0f052ab7cc3b883f22fc84 https://doi.org/https://doi.org/10.48550/arxiv.1904.08935

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: