Machine learning has demonstrated success in clinical risk prediction modeling with complex electronic health record (EHR) data. However, the evolving nature of clinical practices can dynamically change the underlying data distribution over time, leading to model performance drift. Adopting an outdated model is potentially risky and may result in unintentional losses. In this paper, we propose a novel Hybrid Adaptive Boosting approach (HA-Boost) for transfer learning. HA-Boost is characterized by the domain similarity-based and class imbalance-based adaptation mechanisms, which simultaneously address two critical limitations of the classical TrAdaBoost algorithm. We validated HA-Boost in predicting hospital-acquired acute kidney injury using real-world longitudinal EHRs data. The experiment results demonstrate that HA-Boost stably outperforms the competing baselines in terms of both Area Under Receiver Operating Characteristic and Area Under Precision-Recall Curve across a 7-year time span. This study has confirmed the effectiveness of transfer learning as a superior model updating approach in a dynamic environment.

BACKGROUND: Drug-drug interaction (DDI) is a serious public health issue. The L1000 database of the LINCS project has collected millions of genome-wide expressions induced by 20,000 small molecular compounds on 72 cell lines. Whether this unified and comprehensive transcriptome data resource can be used to build a better DDI prediction model is still unclear. Therefore, we developed and validated a novel deep learning model for predicting DDI using 89,970 known DDIs extracted from the DrugBank database (version 5.1.4). RESULTS: The proposed model consists of a graph convolutional autoencoder network (GCAN) for embedding drug-induced transcriptome data from the L1000 database of the LINCS project; and a long short-term memory (LSTM) for DDI prediction. Comparative evaluation of various machine learning methods demonstrated the superior performance of our proposed model for DDI prediction. Many of our predicted DDIs were revealed in the latest DrugBank database (version 5.1.7). In the case study, we predicted drugs interacting with sulfonylureas to cause hypoglycemia and drugs interacting with metformin to cause lactic acidosis, and showed both to induce effects on the proteins involved in the metabolic mechanism in vivo. CONCLUSIONS: The proposed deep learning model can accelerate the discovery of new DDIs. It can support future clinical research for safer and more effective drug co-prescription.

Location is an important distinguishing information for instance segmentation. In this paper, we propose a novel model, called Location Sensitive Network (LSNet), for human instance segmentation. LSNet integrates instance-specific location information into one-stage segmentation framework. Specifically, in the segmentation branch, Pose Attention Module (PAM) encodes the location information into the attention regions through coordinates encoding. Based on the location information provided by PAM, the segmentation branch is able to effectively distinguish instances in feature-level. Moreover, we propose a combination operation named Keypoints Sensitive Combination (KSCom) to utilize the location information from multiple sampling points. These sampling points construct the points representation for instances via human keypoints and random points. Human keypoints provide the spatial locations and semantic information of the instances, and random points expand the receptive fields. Based on the points representation for each instance, KSCom effectively reduces the mis-classified pixels. Our method is validated by the experiments on public datasets. LSNet-5 achieves 56.2 mAP at 18.5 FPS on COCOPersons. Besides, the proposed method is significantly superior to its peers in the case of severe occlusion.