BACKGROUND Machine learning-based Alzheimer's detection using natural language processing has drawn increasing attention because of its low cost compared with traditional methods. However, most of these models are black-boxes, and the decision mechanisms of the AI are obscure. In some fields like medicine, this obscurity gets in the way of widespread adoption. This has led to the development of a new class of techniques that are generally referred to as explainable AI (XAI). One approach to this problem is counter-factual explanations which answer “what if” questions like “What would have happened to Y, had I not done X?”. OBJECTIVE This study aims to improve the transparency of a the-state-of-art language-based Alzheimer’s disease (AD) detection model and discover linguistic biomarkers that are indicative of AD and hence can be used as tools for automated diagnosis of AD. METHODS In this paper, a new explainable artificial intelligence (XAI) method is proposed and named one-intervention counterfactual explanation (OICE). This method works on the state-of-the-art language-based, deep learning method for AD detection and provides an explanation of that method. The proposed OICE incorporates causal factors among the features used in the detection of AD, to provide more transparency of the AI’s decision. This is in contrast to conventional counterfactual explanation methods which do not incorporate causal mechanisms. An understanding of causal factors can go beyond mere statistical correlation to provide a better understanding of the underlying physical phenomenon. The proposed OICE generates counterfactual explanations from a predefined deep-based structural causal model (SCM). The proposed method generated explanations of the AI’s decision by only intervening on one feature at a time. Since OICE provides explanations for individual samples, we then analyze the counterfactual explanations statistically and define some metrics to quantify the effect of every feature. RESULTS We find 11 language level biomarkers for Alzheimer’s disease detection such as adverb, pronoun, noun, preposition, etc. Previous work in psychology and NLP points out adverbs, pronouns, and nouns as potential biomarkers. Our study concurs. We also find new biomarkers that were not reported in previous studies, such as preposition, predeterminer, etc. Our results also reveal how these biomarkers are involved in the diagnostic process from a causal perspective. For example, an on-average 20.2% increase in predeterminer, causes determiner, verb (present particle), and grammatical particles change, resulting in flipping in the diagnosis from control to Alzheimer’s disease. This implies that predeterminer is potentially a strong indicator of the individual’s health and can function as a strong biomarker. CONCLUSIONS Our findings show consistency with previous works in psychology and natural language processing (NLP). Additionally, we offer a new explanation about how intervening a feature can affect the model's decisions using the pre-defined SCM.
Background Recently, rich computational methods that use deep learning or machine learning have been developed using linguistic biomarkers for the diagnosis of early-stage Alzheimer disease (AD). Moreover, some qualitative and quantitative studies have indicated that certain part-of-speech (PoS) features or tags could be good indicators of AD. However, there has not been a systematic attempt to discover the underlying relationships between PoS features and AD. Moreover, there has not been any attempt to quantify the relative importance of PoS features in detecting AD. Objective Our goal was to disclose the underlying relationship between PoS features and AD, understand whether PoS features are useful in AD diagnosis, and explore which PoS features play a vital role in the diagnosis. Methods The DementiaBank, containing 1049 transcripts from 208 patients with AD and 243 transcripts from 104 older control individuals, was used. A total of 27 PoS features were extracted from each record. Then, the relationship between AD and each of the PoS features was explored. A transformer-based deep learning model for AD prediction using PoS features was trained. Then, a global explainable artificial intelligence method was proposed and used to discover which PoS features were the most important in AD diagnosis using the transformer-based predictor. A global (model-level) feature importance measure was derived as a summary from the local (example-level) feature importance metric, which was obtained using the proposed causally aware counterfactual explanation method. The unique feature of this method is that it considers causal relations among PoS features and can, hence, preclude counterfactuals that are improbable and result in more reliable explanations. Results The deep learning–based AD predictor achieved an accuracy of 92.2% and an F1-score of 0.955 when distinguishing patients with AD from healthy controls. The proposed explanation method identified 12 PoS features as being important for distinguishing patients with AD from healthy controls. Of these 12 features, 3 (25%) have been identified by other researchers in previous works in psychology and natural language processing. The remaining 75% (9/12) of PoS features have not been previously identified. We believe that this is an interesting finding that can be used in creating tests that might aid in the diagnosis of AD. Note that although our method is focused on PoS features, it should be possible to extend it to more types of features, perhaps even those derived from other biomarkers, such as syntactic features. Conclusions The high classification accuracy of the proposed deep learner indicates that PoS features are strong clues in AD diagnosis. There are 12 PoS features that are strongly tied to AD, and because language is a noninvasive and potentially cheap method for detecting AD, this work shows some promising directions in this field.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.