Addressing Noise and Skewness in Interpretable Health-Condition Assessment by Learning Model Confidence

Abstract

Assessing the health condition has a wide range of applications in healthcare, military, aerospace, and industrial fields. Nevertheless, traditional feature-engineered techniques involve manual feature extraction, which are too cumbersome to adapt to the changes caused by the development of sensor network technology. Recently, deep-learning-based methods have achieved initial success in health-condition assessment research, but insufficient considerations for problems such as class skewness, noisy segments, and result interpretability make it difficult to apply them to real-world applications. In this paper, we propose a K-margin-based Interpretable Learning approach for health-condition assessment. In detail, a skewness-aware RCR-Net model is employed to handle problems of class skewness. Furthermore, we present a diagnosis model based on K-margin to automatically handle noisy segments by naturally exploiting expected consistency among the segments associated with each record. Additionally, a knowledge-directed interpretation method is presented to learn domain knowledge-level features automatically without the help of human experts which can be used as an interpretable decision-making basis. Finally, through experimental validation in the field of both medical and aerospace, the proposed method has a better generality and high efficiency with 0.7974 and 0.8005 F1 scores, which outperform all state-of-the-art deep learning methods for health-condition assessment task by 3.30% and 2.99%, respectively.

Artificial-Intelligence-Enhanced Mobile System for Cardiovascular Health Management

“…AI techniques have recently shown significant potential in cardiology [ 22 , 23 , 24 , 25 , 26 , 27 ] owing to their ability to automatically learn effective features from data without the help of domain experts. When focusing on deep learning methods applying ECG data, various architectures have been proposed for disease detection [ 15 , 17 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 , 38 ], sleep staging [ 39 , 40 ], and biometric identification [ 41 , 42 , 43 , 44 ], among others (see a recent survey in [ 22 ]).…”

Application of Explainable AI (Xai) For Anomaly Detection and Prognostic of Gas Turbines with Uncertainty Quantification.

“…This section elaborates some of the former PHM-XAI articles available. In presentation order: Interpretable model [17], tree-based [18], knowledge & rule-based [19], Logic Analysis of Data (LAD) [20], feature extraction-based [21], filter-based [22], cluster-based [23], attention-based [24], model-agnostic explainability [25] and Layer-Wise Relevance Propagation (LRP) [26].…”

“…A K-margin-based intErpretable lEarNing (KEEN) is presented in [19] for interpretable aircraft structural damage diagnosis. This framework consists of a Residual Convolution Recurrent Neural Network (RCR-Net), a K-margin diagnostic method and a knowledge-directed interpretation approach.…”

Explainable Artificial Intelligence for Anomaly Detection and Prognostic of Gas Turbines using Uncertainty Quantification with Sensor-Related Data

“…Next, we elaborate some literature on PHM-XAI. In presentation order: Interpretable model [19], tree-based [20], knowledge and rule-based [21], logic analysis of data (LAD) [22], feature extraction-based [23], filter-based [24], cluster-based [23], attention-based [26], model-agnostic explainability [27], and layer-wise relevance propagation (LRP) [28].…”

“…A K-margin-based interpretable learning (KEEN) is presented in [21] for interpretable aircraft structural damage diagnostic. This framework consists of a residual convolution recurrent neural network (RCR-net), a K-margin diagnostic method and a knowledge-directed interpretation approach.…”