Alexander Brenner , Felix Knispel , Florian P. Fischer , Peter Rossmanith , Yvonne Weber , Henner Koch , Rainer Röhrig , Julian Varghese , Ekaterina Kutafina
{"title":"Concept-based AI interpretability in physiological time-series data: Example of abnormality detection in electroencephalography","authors":"Alexander Brenner , Felix Knispel , Florian P. Fischer , Peter Rossmanith , Yvonne Weber , Henner Koch , Rainer Röhrig , Julian Varghese , Ekaterina Kutafina","doi":"10.1016/j.cmpb.2024.108448","DOIUrl":null,"url":null,"abstract":"<div><h3>Background and Objective</h3><div>Despite recent performance advancements, deep learning models are not yet adopted in clinical practice on a wide scale. The intrinsic intransparency of such systems is commonly cited as one major reason for this reluctance. This has motivated methods that aim to provide explanations of model functioning. Known limitations of feature-based explanations have led to an increased interest in concept-based interpretability. <em>Testing with Concept Activation Vectors</em> (TCAV) employs human-understandable, abstract concepts to explain model behavior. The method has previously been applied to the medical domain in the context of electronic health records, retinal fundus images and magnetic resonance imaging.</div></div><div><h3>Methods</h3><div>We explore the usage of TCAV for building interpretable models on physiological time series, using an example of abnormality detection in electroencephalography (EEG). For this purpose, we adopt the XceptionTime model, which is suitable for multi-channel physiological data of variable sizes. The model provides state-of-the-art performance on raw EEG data and is publically available. We propose and test several ideas regarding concept definition through metadata mining, using additional labeled EEG data and extracting interpretable signal characteristics in the form of frequencies. By including our own hospital data with analog labeling, we further evaluate the robustness of our approach.</div></div><div><h3>Results</h3><div>The tested concepts show a TCAV score distribution that is in line with the clinical expectations, i.e. concepts known to have strong links with EEG pathologies (such as epileptiform discharges) received higher scores than the neutral concepts (e.g. sex). The scores were consistent across the applied concept generation strategies.</div></div><div><h3>Conclusions</h3><div>TCAV has the potential to improve interpretability of deep learning applied to multi-channel signals as well as to detect possible biases in the data. Still, further work on developing the strategies for concept definition and validation on clinical physiological time series is needed to better understand how to extract clinically relevant information from the concept sensitivity scores.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"257 ","pages":"Article 108448"},"PeriodicalIF":4.9000,"publicationDate":"2024-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Computer methods and programs in biomedicine","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0169260724004413","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS","Score":null,"Total":0}
引用次数: 0
Abstract
Background and Objective
Despite recent performance advancements, deep learning models are not yet adopted in clinical practice on a wide scale. The intrinsic intransparency of such systems is commonly cited as one major reason for this reluctance. This has motivated methods that aim to provide explanations of model functioning. Known limitations of feature-based explanations have led to an increased interest in concept-based interpretability. Testing with Concept Activation Vectors (TCAV) employs human-understandable, abstract concepts to explain model behavior. The method has previously been applied to the medical domain in the context of electronic health records, retinal fundus images and magnetic resonance imaging.
Methods
We explore the usage of TCAV for building interpretable models on physiological time series, using an example of abnormality detection in electroencephalography (EEG). For this purpose, we adopt the XceptionTime model, which is suitable for multi-channel physiological data of variable sizes. The model provides state-of-the-art performance on raw EEG data and is publically available. We propose and test several ideas regarding concept definition through metadata mining, using additional labeled EEG data and extracting interpretable signal characteristics in the form of frequencies. By including our own hospital data with analog labeling, we further evaluate the robustness of our approach.
Results
The tested concepts show a TCAV score distribution that is in line with the clinical expectations, i.e. concepts known to have strong links with EEG pathologies (such as epileptiform discharges) received higher scores than the neutral concepts (e.g. sex). The scores were consistent across the applied concept generation strategies.
Conclusions
TCAV has the potential to improve interpretability of deep learning applied to multi-channel signals as well as to detect possible biases in the data. Still, further work on developing the strategies for concept definition and validation on clinical physiological time series is needed to better understand how to extract clinically relevant information from the concept sensitivity scores.
期刊介绍:
To encourage the development of formal computing methods, and their application in biomedical research and medical practice, by illustration of fundamental principles in biomedical informatics research; to stimulate basic research into application software design; to report the state of research of biomedical information processing projects; to report new computer methodologies applied in biomedical areas; the eventual distribution of demonstrable software to avoid duplication of effort; to provide a forum for discussion and improvement of existing software; to optimize contact between national organizations and regional user groups by promoting an international exchange of information on formal methods, standards and software in biomedicine.
Computer Methods and Programs in Biomedicine covers computing methodology and software systems derived from computing science for implementation in all aspects of biomedical research and medical practice. It is designed to serve: biochemists; biologists; geneticists; immunologists; neuroscientists; pharmacologists; toxicologists; clinicians; epidemiologists; psychiatrists; psychologists; cardiologists; chemists; (radio)physicists; computer scientists; programmers and systems analysts; biomedical, clinical, electrical and other engineers; teachers of medical informatics and users of educational software.