RESUMO
Alzheimer's disease (AD) stands as the prevalent progressive neurodegenerative disease, precipitating cognitive impairment and even memory loss. Amyloid biomarkers have been extensively used in the diagnosis of AD. However, amyloid proteins offer limited information about the disease process and accurate diagnosis depends on the presence of a substantial accumulation of amyloid deposition which significantly impedes the early screening of AD. In this study, we have combined plasma proteomics with an ensemble learning model (CatBoost) to develop a cost-effective and non-invasive diagnostic method for AD. A longitudinal panel has been identified that can serve as reliable biomarkers across the entire progression of AD. Simultaneously, we have developed a neural network algorithm that utilizes plasma proteins to detect stages of Alzheimer's disease. Based on the developed longitudinal panel, the CatBoost model achieved an area under the operating curve of at least 0.90 in distinguishing mild cognitive impairment from cognitively normal. The neural network model was utilized for the detection of three stages of AD, and the results demonstrated that the neural network model exhibited an accuracy as high as 0.83, surpassing that of the traditional machine learning model.
Assuntos
Doença de Alzheimer , Biomarcadores , Diagnóstico Precoce , Aprendizado de Máquina , Redes Neurais de Computação , Proteoma , Doença de Alzheimer/sangue , Doença de Alzheimer/diagnóstico , Humanos , Idoso , Biomarcadores/sangue , Masculino , Feminino , Disfunção Cognitiva/diagnóstico , Disfunção Cognitiva/sangue , Proteômica/métodos , Idoso de 80 Anos ou maisRESUMO
Single-cell omics techniques have made it possible to analyze individual cells in biological samples, providing us with a more detailed understanding of cellular heterogeneity and biological systems. Accurate identification of cell types is critical for single-cell RNA sequencing (scRNA-seq) analysis. However, scRNA-seq data are usually high dimensional and sparse, posing a great challenge to analyze scRNA-seq data. Existing cell-type annotation methods are either constrained in modeling scRNA-seq data or lack consideration of long-term dependencies of characterized genes. In this work, we developed a Transformer-based deep learning method, scSwinFormer, for the cell-type annotation of large-scale scRNA-seq data. Sequence modeling of scRNA-seq data is performed using the smooth gene embedding module, and then, the potential dependencies of genes are captured by the self-attention module. Subsequently, the global information inherent in scRNA-seq data is synthesized using the Cell Token, thereby facilitating accurate cell-type annotation. We evaluated the performance of our model against current state-of-the-art scRNA-seq cell-type annotation methods on multiple real data sets. ScSwinFormer outperforms the current state-of-the-art scRNA-seq cell-type annotation methods in both external and benchmark data set experiments.