PreCanCell: An ensemble learning algorithm for predicting cancer and non-cancer cells from single-cell transcriptomes

scMalignantFinder distinguishes malignant cells in single-cell and spatial transcriptomics by leveraging cancer signatures

Single-cell RNA sequencing (scRNA-seq) is a powerful tool for characterizing tumor heterogeneity, yet accurately identifying malignant cells remains challenging. Here, we propose scMalignantFinder, a machine learning tool specifically designed to distinguish malignant cells from their normal counterparts using a data- and knowledge-driven strategy. To develop the tool, multiple cancer datasets were collected, and the initially annotated malignant cells were calibrated using nine carefully curated pan-cancer gene signatures, resulting in over 400,000 single-cell transcriptomes for training. The union of differentially expressed genes across datasets was taken as the features for model construction to comprehensively capture tumor transcriptional diversity. scMalignantFinder outperformed existing automated methods across two gold-standard and eleven patient-derived scRNA-seq datasets. The capability to predict malignancy probability empowers scMalignantFinder to capture dynamic characteristics during tumor progression. Furthermore, scMalignantFinder holds the potential to annotate malignant regions in tumor spatial transcriptomics. Overall, we provide an efficient tool for detecting heterogeneous malignant cell populations.

Inferring tumor purity using multi-omics data based on a uniform machine learning framework MoTP

Existing algorithms for assessing tumor purity are limited to a single omics data, such as gene expression, somatic copy number variations, somatic mutations, and DNA methylation. Here we proposed the machine learning Multi-omics Tumor Purity prediction (MoTP) algorithm to estimate tumor purity based on multiple types of omics data. MoTP utilizes the Bayesian Regularized Neural Networks as the prediction algorithm, and Consensus Tumor Purity Estimates as labels. We trained MoTP using multi-omics data (mRNA, microRNA, long non-coding RNA, and DNA methylation) across 21 TCGA solid cancer types. By testing MoTP in TCGA validation sets, TCGA test sets, and eight datasets outside the TCGA cancer cohorts, we showed that although MoTP could achieve excellent performance in predicting tumor purity based on a single omics data type, the integration of multiple single omics data-based predictions can enhance the prediction performance. Moreover, we demonstrated the robustness of MoTP by testing it in datasets with Gaussian noise and feature missing. Benchmark analysis showed that MoTP outperformed most established tumor purity prediction algorithms, and that it required less running time and computational resource to fulfill the predictive task. Thus, MoTP would be an attractive option for computational tumor purity inference.

Multi-modal feature fusion with multi-head self-attention for epileptic EEG signals

It is important to classify electroencephalography (EEG) signals automatically for the diagnosis and treatment of epilepsy. Currently, the dominant single-modal feature extraction methods cannot cover the information of different modalities, resulting in poor classification performance of existing methods, especially the multi-classification problem. We proposed a multi-modal feature fusion (MMFF) method for epileptic EEG signals. First, the time domain features were extracted by kernel principal component analysis, the frequency domain features were extracted by short-time Fourier extracted transform, and the nonlinear dynamic features were extracted by calculating sample entropy. On this basis, the features of these three modalities were interactively learned through the multi-head self-attention mechanism, and the attention weights were trained simultaneously. The fused features were obtained by combining the value vectors of feature representations, while the time, frequency, and nonlinear dynamics information were retained to screen out more representative epileptic features and improve the accuracy of feature extraction. Finally, the feature fusion method was applied to epileptic EEG signal classifications. The experimental results demonstrated that the proposed method achieves a classification accuracy of 92.76 ± 1.64% across the five-category classification task for epileptic EEG signals. The multi-head self-attention mechanism promotes the fusion of multi-modal features and offers an efficient and novel approach for diagnosing and treating epilepsy.

Artificial Intelligence in battling infectious diseases: A transformative role

It is widely acknowledged that infectious diseases have wrought immense havoc on human society, being regarded as adversaries from which humanity cannot elude. In recent years, the advancement of Artificial Intelligence (AI) technology has ushered in a revolutionary era in the realm of infectious disease prevention and control. This evolution encompasses early warning of outbreaks, contact tracing, infection diagnosis, drug discovery, and the facilitation of drug design, alongside other facets of epidemic management. This article presents an overview of the utilization of AI systems in the field of infectious diseases, with a specific focus on their role during the COVID-19 pandemic. The article also highlights the contemporary challenges that AI confronts within this domain and posits strategies for their mitigation. There exists an imperative to further harness the potential applications of AI across multiple domains to augment its capacity in effectively addressing future disease outbreaks.