Lung cancer is one of the most harmful malignant tumors to human health. The accurate judgment of the pathological type of lung cancer is vital for treatment. Traditionally, the pathological type of lung cancer requires a histopathological examination to determine, which is invasive and time consuming. In this work, a novel residual neural network is proposed to identify the pathological type of lung cancer via CT images. Due to the low amount of CT images in practice, we explored a medical-to-medical transfer learning strategy. Specifically, a residual neural network is pre-trained on public medical images dataset luna16, and then fine-tuned on our intellectual property lung cancer dataset collected in Shandong Provincial Hospital. Data experiments show that our method achieves 85.71% accuracy in identifying pathological types of lung cancer from CT images and outperforming other models trained with 2054 labels. Our method performs better than AlexNet, VGG16 and DenseNet, which provides an efficient, non-invasive detection tool for pathological diagnosis.

The properties of the drug may be altered by the combination, which may cause unexpected drug–drug interactions (DDIs). Prediction of DDIs provides combination strategies of drugs for systematic and effective treatment. In most of deep learning-based methods for predicting DDI, encoded information about the drugs is insufficient in some extent, which limits the performances of DDIs prediction. In this work, we propose a novel attention-mechanism-based multidimensional feature encoder for DDIs prediction, namely attention-based multidimensional feature encoder (AMDE). Specifically, in AMDE, we encode drug features from multiple dimensions, including information from both Simplified Molecular-Input Line-Entry System sequence and atomic graph of the drug. Data experiments are conducted on DDI data set selected from Drugbank, involving a total of 34 282 DDI relationships with 17 141 positive DDI samples and 17 141 negative samples. Experimental results show that our AMDE performs better than some state-of-the-art baseline methods, including Random Forest, One-Dimension Convolutional Neural Networks, DeepDrug, Long Short-Term Memory, Seq2seq, Deepconv, DeepDDI, Graph Attention Networks and Knowledge Graph Neural Networks. In practice, we select a set of 150 drugs with 3723 DDIs, which are never appeared in training, validation and test sets. AMDE performs well in DDIs prediction task, with AUROC and AUPRC 0.981 and 0.975. As well, we use Torasemide (DB00214) as an example and predict the most likely drug to interact with it. The top 15 scores all have been reported with clear interactions in literatures.

Locating diseases precisely from medical images, like ultrasonic and CT images, have been one of the most challenging problems in medical image analysis. In recent years, the vigorous development of deep learning models have greatly improved the accuracy in disease location on medical images. However, there are few artificial intelligent methods for identifying cholelithiasis and classifying gallstones on CT images, since no open source CT images dataset of cholelithiasis and gallstones is available for training the models and verifying their performance. In this paper, we build up the first medical image dataset of cholelithiasis by collecting 223846 CT images with gallstone of 1369 patients. With these CT images, a neural network is trained to “pick up” CT images of high quality as training set, and then a novel Yolo neural network, named Yolov3-arch neural network, is proposed to identify cholelithiasis and classify gallstones on CT images. Identification and classification accuracies are obtained by 10-fold cross-validations. It is obtained that our Yolov3-arch model is with average accuracy 92.7% in identifying granular gallstones and average accuracy 80.3% in identifying muddy gallstones. This achieves 3.5% and 8% improvements in identifying granular and muddy gallstones to general Yolo v3 model, respectively. Also, the average cholelithiasis identifying accuracy is improved to 86.50% from 80.75%. Meanwhile, our method can reduce the misdiagnosis rate of negative samples by the object detection model.

Multi-drug combinations for the treatment of complex diseases are gradually becoming an important treatment, and this type of treatment can take advantage of the synergistic effects among drugs. However, drug–drug interactions (DDIs) are not just all beneficial. Accurate and rapid identifications of the DDIs are essential to enhance the effectiveness of combination therapy and avoid unintended side effects. Traditional DDIs prediction methods use only drug sequence information or drug graph information, which ignores information about the position of atoms and edges in the spatial structure. In this paper, we propose Molormer, a method based on a lightweight attention mechanism for DDIs prediction. Molormer takes the two-dimension (2D) structures of drugs as input and encodes the molecular graph with spatial information. Besides, Molormer uses lightweight-based attention mechanism and self-attention distilling to process spatially the encoded molecular graph, which not only retains the multi-headed attention mechanism but also reduces the computational and storage costs. Finally, we use the Siamese network architecture to serve as the architecture of Molormer, which can make full use of the limited data to train the model for better performance and also limit the differences to some extent between networks dealing with drug features. Experiments show that our proposed method outperforms state-of-the-art methods in Accuracy, Precision, Recall and F1 on multi-label DDIs dataset. In the case study section, we used Molormer to make predictions of new interactions for the drugs Aliskiren, Selexipag and Vorapaxar and validated parts of the predictions. Code and models are available at https://github.com/IsXudongZhang/Molormer.

Recent years have seen tremendous success in the design of novel drug molecules through deep generative models. Nevertheless, existing methods only generate drug-like molecules, which require additional structural optimization to be developed into actual drugs. In this study, a deep learning method for generating target-specific ligands was proposed. This method is useful when the dataset for target-specific ligands is limited. Deep learning methods can extract and learn features (representations) in a data-driven way with little or no human participation. Generative pretraining (GPT) was used to extract the contextual features of the molecule. Three different protein-encoding methods were used to extract the physicochemical properties and amino acid information of the target protein. Protein-encoding and molecular sequence information are combined to guide molecule generation. Transfer learning was used to fine-tune the pretrained model to generate molecules with better binding ability to the target protein. The model was validated using three different targets. The docking results show that our model is capable of generating new molecules with higher docking scores for the target proteins.

Histological analysis to tissue samples is elemental for diagnosing the risk and severity of ovarian cancer. The commonly used Hematoxylin and Eosin (H&E) staining method involves complex steps and strict requirements, which would seriously impact the research of histological analysis of the ovarian cancer. Virtual histological staining by the Generative Adversarial Network (GAN) provides a feasible way for these problems, yet it is still a challenge of using deep learning technology since the amounts of data available are quite limited for training. Based on the idea of GAN, we propose a weakly supervised learning method to generate autofluorescence images of unstained ovarian tissue sections corresponding to H&E staining sections of ovarian tissue. Using the above method, we constructed the supervision conditions for the virtual staining process, which makes the image quality synthesized in the subsequent virtual staining stage more perfect. Through the doctors’ evaluation of our results, the accuracy of ovarian cancer unstained fluorescence image generated by our method reached 93%. At the same time, we evaluated the image quality of the generated images, where the FID reached 175.969, the IS score reached 1.311, and the MS reached 0.717. Based on the image-to-image translation method, we use the data set constructed in the previous step to implement a virtual staining method that is accurate to tissue cells. The accuracy of staining through the doctor’s assessment reached 97%. At the same time, the accuracy of visual evaluation based on deep learning reached 95%.