Arrhythmia accounts for more than 80% of sudden cardiac death, and its incidence rate has increased rapidly recently. Nowadays, many studies have applied artificial intelligence (AI) methods to arrhythmia detection. Deep learning approaches can improve model performance but lack the capability to mine the relationship between the electrocardiogram (ECG) samples and final results. Furthermore, most arrhythmia diagnosis methods aim at only multiclassification problems. However, in many cases, an ECG recording may have more than one arrhythmia disease. In this article, we develop an accurate and interpretable model for multilabel ECG signals, which is called dual-level attentional convolutional long short-term memory neural network (DLA-CLSTM). Convolutional layers, bidirectional long short-term memory layers, and attention mechanisms are used for extracting beat- and rhythm-level information. Notably, the attention mechanism helps the model be more interpretable. In addition, we design a weighted multilabel classifier to address multilabel and imbalance problems. Finally, the MIT-BIH Arrhythmia Database and the 1st China Physiological Signal Challenge dataset are utilized to evaluate DLA-CLSTM. Compared with existing methods, DLA-CLSTM can improve the accuracy by 22.50% and the F1-macro by 20.51% on average. Further analysis shows that the areas focused by DLA-CLSTM match the clinical judgment basis to some extent. Collectively, DLA-CLSTM has good potential for assisting clinicians in real applications.

Background: Long-term monitoring of Electrocardiogram (ECG) recordings is crucial to diagnose arrhythmias. Clinicians can find it challenging to diagnose arrhythmias, and this is a particular issue in more remote and underdeveloped areas. The development of digital ECG and AI methods could assist clinicians who need to diagnose arrhythmias outside of the hospital setting. Methods: We constructed a large-scale Chinese ECG benchmark dataset using data from 272,753 patients collected from January 2017 to December 2021. The dataset contains ECG recordings from all common arrhythmias present in the Chinese population. Several experienced cardiologists from Shanghai First People’s Hospital labeled the dataset. We then developed a deep learning-based multi-label interpretable diagnostic model from the ECG recordings. We utilized Accuracy, F1 score and AUC-ROC to compare the performance of our model with that of the cardiologists, as well as with six comparison models, using testing and hidden data sets. Results: The results show that our approach achieves an F1 score of 83.51%, an average AUC ROC score of 0.977, and 93.74% mean accuracy for 6 common arrhythmias. Results from the hidden dataset demonstrate the performance of our approach exceeds that of cardiologists. Our approach also highlights the diagnostic process. Conclusions: Our diagnosis system has superior diagnostic performance over that of clinicians. It also has the potential to help clinicians rapidly identify abnormal regions on ECG recordings, thus improving efficiency and accuracy of clinical ECG diagnosis in China. This approach could therefore potentially improve the productivity of out-of-hospital ECG diagnosis and provides a promising prospect for telemedicine.

In the actual scenario, it is difficult for an automatic diagnosis system to detect each heartbeat when facing new patients. In this article, a novel active dual-scale residual convolutional long short-term memory neural network is proposed to classify heartbeats interactively and automatically. First, each heartbeat and electrocardiogram segment including current heartbeat are combined to build dual-scale input information. In addition, convolution layers, residual block, and bidirectional long short memory are integrated to extract features from dual-scale inputs. Furthermore, active learning approach with random breaking-ties selection strategy is introduced to choose the most representative unlabeled samples for labeling and fine-tune the trained model with these representative labeled samples. The experimental results show that the proposed method uses only 5.14% of the additional data set and improves the model accuracy and macro F1 score by 13.23% and 35.62%, respectively. The proposed structure has good representative ability and our selection strategy can reduce the workload of experts and improve the performance greatly. Thus, the proposed interactive and incremental method has good potential in real applications.

Reinforcement learning (RL) still suffers from the problem of sample inefficiency and struggles with the exploration issue, particularly in situations with long-delayed rewards, sparse rewards, and deep local optimum. Recently, learning from demonstration (LfD) paradigm was proposed to tackle this problem. However, these methods usually require a large number of demonstrations. In this study, we present a sample efficient teacher-advice mechanism with Gaussian process (TAG) by leveraging a few expert demonstrations. In TAG, a teacher model is built to provide both an advice action and its associated confidence value. Then, a guided policy is formulated to guide the agent in the exploration phase via the defined criteria. Through the TAG mechanism, the agent is capable of exploring the environment more intentionally. Moreover, with the confidence value, the guided policy can guide the agent precisely. Also, due to the strong generalization ability of Gaussian process, the teacher model can utilize the demonstrations more effectively. Therefore, substantial improvement in performance and sample efficiency can be attained. Considerable experiments on sparse reward environments demonstrate that the TAG mechanism can help typical RL algorithms achieve significant performance gains. In addition, the TAG mechanism with soft actor-critic algorithm (TAG-SAC) attains the state-of-the-art performance over other LfD counterparts on several delayed reward and complicated continuous control environments.

Existing prognostic models are useful for estimating the prognosis of lung adenocarcinoma patients, but there remains room for improvement. In the current study, we developed a deep learning model based on histopathological images to predict the recurrence risk of lung adenocarcinoma patients. The efficiency of the model was then evaluated in independent multicenter cohorts. The model defined high- and low-risk groups successfully stratified prognosis of the entire cohort. Moreover, multivariable Cox analysis identified the model defined risk groups as an independent predictor for disease-free survival. Importantly, combining TNM stage with the established model helped to distinguish subgroups of patients with high-risk stage II and stage III disease who are highly likely to benefit from adjuvant chemotherapy. Overall, our study highlights the significant value of the constructed model to serve as a complementary biomarker for survival stratification and adjuvant therapy selection for lung adenocarcinoma patients after resection.