This paper introduces a dual-modal early cognitive impairment detection system based on autobiographical memory (AM) tests, and our approach is to automatically extract pre-defined acoustic features and self-designed embeddings to enhance linguistic representation of the spontaneous speech data. By integrating dual-modal data, we effectively enrich the features that aid in model learning, especially addressing the subtle symptoms exhibited by individuals with mild cognitive impairment (MCI), an intermediate stage between healthy individuals and those with Alzheimer's disease (AD). To account for spontaneous speech's unstructured and implicit nature, two additional embeddings, namely, speaker embedding and conversation embedding, are introduced to augment the information available for model learning, thus enriching the feature set for improving the model accuracy. The proposed dual-modal approach is tested on a self-collected Chinese spontaneous speech dataset due to the limited unstructured speech open-access dataset for MCI detection. The system's effectiveness is evaluated through a series of experiments, including ablation studies, to determine the impact of each module on overall performance. The proposed system achieved an average accuracy of 78% in detecting MCI, demonstrating its comparative effectiveness. Enhancements in our system are achieved by integrating a directional encoder tailored to capture temporal information across sequential visits. This addition leads to a 3% increase in detection accuracy within a subset of participants who have undergone multiple AM test sessions. Implementing such a longitudinal approach in analyzing unstructured speech data for MCI detection taps into a relatively underexplored area of research, offering new insights.

Traditional text classification requires thousands of annotated data or an additional Neural Machine Translation (NMT) system, which are expensive to obtain in real applications. This paper presents a Contrast-Enhanced Semi-supervised Text Classification (CEST) framework under label-limited settings without incorporating any NMT systems. We propose a certainty-driven sample selection method and a contrast-enhanced similarity graph to utilize data more efficiently in self-training, alleviating the annotation-starving problem. The graph imposes a smoothness constraint on the unlabeled data to improve the coherence and the accuracy of pseudo-labels. Moreover, CEST formulates the training as a “learning from noisy labels” problem and performs the optimization accordingly. A salient feature of this formulation is the explicit suppression of the severe error propagation problem in conventional semi-supervised learning. With solely 30 labeled data per class for both training and validation dataset, CEST outperforms the previous state-of-the-art algorithms by 2.11% accuracy and only falls within the 3.04% accuracy range of fully-supervised pre-training language model fine-tuning on thousands of labeled data.

Alzheimer’s disease (AD) and other types of dementia have become a public health priority worldwide. To lessen the burden of AD diagnosis, an automatic screening system that can be deployed in large-scale and cost-efficient screening methods will be needed. This paper presents a speech assessment system for cognitive impairment detection, detecting whether elders have AD or suffer from mild cognitive impairment (MCI) based on their audio recordings taken from neuropsychological tests. The audio waveform first is transformed to Mel-spectrogram and done the downsampling. With the combination of Transformer and convolutional neural network (CNN) architecture, we can do the feature extraction and get a better representation for the classifier. We conducted experiments on 120 subjects with a balanced distribution of ordinary aging, MCI, and AD patients to validate our study. Our experiments achieve an accuracy of 91% and 79% for classifying groups of AD and MCI from ordinary aging people, respectively.

As the elderly population grows globally, health-care systems face a burden from the rise in Alzheimer’s patients due to an increase in demand for early diagnosis. Therefore, more people have started focusing on developing systems helping doctors diagnose Alzheimer’s, such as cognitive impairment detection systems. This paper presents a contrast-enhanced automatic cognitive impairment screening system combining paused-encoder based on the automatic transcription. We use the pre-trained automatic speech recognition model and adapt it to generate transcripts of the elderly’s speech. The pattern of pauses in speech is a commonly-studied acoustic feature since it can provide additional information besides the semantic information for the model prediction. The back-translation with contrastive learning is used to improve the encoded model further. The model also fine-tunes with the pause-encoded transcriptions to detect the cognitive impairment. Our result shows excellent performance with an accuracy of 81% in detecting Alzheimer’s disease. Also, the accuracy is acceptable on a more challenging task of detecting mild cognitive impairment, the middle stage between healthy and Alzheimer’s. In addition to the outperforming performance, our system is fully automatic and can be used easily.

We propose a Progressive Email Classifier (PEC) for high-speed classification of message patterns that are commonly associated with unsolicited bulk email (UNBE). PEC is designed to operate at the network access point, the ingress between the Internet Service Provider (ISP) and the enterprise network; so that a surge of UNBE containing fresh patterns can be detected before they spread into the enterprise network. A real-time scoreboard keeps track of detected feature instances (FI) based on a scoring and aging engine, until they are considered either from valid or UNBE sources. A FI of a valid email is discarded, but an anomalous one is passed to a blacklist to control (e.g., block or defer) subsequent emails containing the FI.