Multimodal transistors as ReLU activation functions in physical neural network classifiers

Neuron signal attenuation activation mechanism for deep learning

Neuron signal activation is at the core of deep learning and broadly impacts science and engineering. Despite growing interest in neuron cell stimulation via amplitude current, the activation mechanism of biological neurons has limited application in deep learning due to the lack of a universal mathematical principle suitable for artificial neural networks. Here, we show how deep learning can go beyond the current learning effects through a newly proposed neuron signal activation mechanism. To achieve this, we report a new cross-disciplinary method for neuron signal attenuation, using the inference of differential equations within generalized linear systems to enhance the efficiency of deep learning. We formulate the mathematical model of the efficient activation function, which we refer to as Attenuation (Ant). Ant can represent higher-order derivatives and stabilize data distributions in deep-learning tasks. We demonstrate the effectiveness, stability, and generalization of Ant on many challenging tasks across various neural network architectures.

Predicting gastric cancer tumor mutational burden from histopathological images using multimodal deep learning

Tumor mutational burden (TMB) is a significant predictive biomarker for selecting patients that may benefit from immune checkpoint inhibitor therapy. Whole exome sequencing is a common method for measuring TMB; however, its clinical application is limited by the high cost and time-consuming wet-laboratory experiments and bioinformatics analysis. To address this challenge, we downloaded multimodal data of 326 gastric cancer patients from The Cancer Genome Atlas, including histopathological images, clinical data and various molecular data. Using these data, we conducted a comprehensive analysis to investigate the relationship between TMB, clinical factors, gene expression and image features extracted from hematoxylin and eosin images. We further explored the feasibility of predicting TMB levels, i.e. high and low TMB, by utilizing a residual network (Resnet)-based deep learning algorithm for histopathological image analysis. Moreover, we developed a multimodal fusion deep learning model that combines histopathological images with omics data to predict TMB levels. We evaluated the performance of our models against various state-of-the-art methods using different TMB thresholds and obtained promising results. Specifically, our histopathological image analysis model achieved an area under curve (AUC) of 0.749. Notably, the multimodal fusion model significantly outperformed the model that relied only on histopathological images, with the highest AUC of 0.971. Our findings suggest that histopathological images could be used with reasonable accuracy to predict TMB levels in gastric cancer patients, while multimodal deep learning could achieve even higher levels of accuracy. This study sheds new light on predicting TMB in gastric cancer patients.

An intelligent diabetes classification and perception framework based on ensemble and deep learning method

Sugar in the blood can harm individuals and their vital organs, potentially leading to blindness, renal illness, as well as kidney and heart diseases. Globally, diabetic patients face an average annual mortality rate of 38%. This study employs Chi-square, mutual information, and sequential feature selection (SFS) to choose features for training multiple classifiers. These classifiers include an artificial neural network (ANN), a random forest (RF), a gradient boosting (GB) algorithm, Tab-Net, and a support vector machine (SVM). The goal is to predict the onset of diabetes at an earlier age. The classifier, developed based on the selected features, aims to enable early diagnosis of diabetes. The PIMA and early-risk diabetes datasets serve as test subjects for the developed system. The feature selection technique is then applied to focus on the most important and relevant features for model training. The experiment findings conclude that the ANN exhibited a spectacular performance in terms of accuracy on the PIMA dataset, achieving a remarkable accuracy rate of 99.35%. The second experiment, conducted on the early diabetes risk dataset using selected features, revealed that RF achieved an accuracy of 99.36%. Based on our experimental results, it can be concluded that our suggested method significantly outperformed baseline machine learning algorithms already employed for diabetes prediction on both datasets.

Split-Gate: Harnessing Gate Modulation Power in Thin-Film Electronics

With the increase in electronic devices across various applications, there is rising demand for selective carrier control. The split-gate consists of a gate electrode divided into multiple parts, allowing for the independent biasing of electric fields within the device. This configuration enables the potential formation of both p- and n-channels by injecting holes and electrons owing to the presence of the two gate electrodes. Applying voltage to the split-gate allows for the control of the Fermi level and, consequently, the barrier height in the device. This facilitates band bending in unipolar transistors and allows ambipolar transistors to operate as if unipolar. Moreover, the split-gate serves as a revolutionary tool to modulate the contact resistance by controlling the barrier height. This approach enables the precise control of the device by biasing the partial electric field without limitations on materials, making it adaptable for various applications, as reported in various types of research. However, the gap length between gates can affect the injection of the electric field for the precise control of carriers. Hence, the design of the gap length is a critical element for the split-gate structure. The primary investigation in this review is the introduction of split-gate technology applied in various applications by using diverse materials, the methods for forming the split-gate in each device, and the operational mechanisms under applied voltage conditions.

Predicting colorectal cancer tumor mutational burden from histopathological images and clinical information using multi-modal deep learning

MOTIVATION

Tumor mutational burden (TMB) is an indicator of the efficacy and prognosis of immune checkpoint therapy in colorectal cancer (CRC). In general, patients with higher TMB values are more likely to benefit from immunotherapy. Though whole-exome sequencing is considered the gold standard for determining TMB, it is difficult to be applied in clinical practice due to its high cost. There are also a few DNA panel-based methods to estimate TMB; however, their detection cost is also high, and the associated wet-lab experiments usually take days, which emphasize the need for faster and cheaper alternatives.

RESULTS

In this study, we propose a multi-modal deep learning model based on a residual network (ResNet) and multi-modal compact bilinear pooling to predict TMB status (i.e. TMB high (TMB_H) or TMB low(TMB_L)) directly from histopathological images and clinical data. We applied the model to CRC data from The Cancer Genome Atlas and compared it with four other popular methods, namely, ResNet18, ResNet50, VGG19 and AlexNet. We tested different TMB thresholds, namely, percentiles of 10%, 14.3%, 15%, 16.3%, 20%, 30% and 50%, to differentiate TMB_H and TMB_L.For the percentile of 14.3% (i.e. TMB value 20) and ResNet18, our model achieved an area under the receiver operating characteristic curve of 0.817 after 5-fold cross-validation, which was better than that of other compared models. In addition, we also found that TMB values were significantly associated with the tumor stage and N and M stages. Our study shows that deep learning models can predict TMB status from histopathological images and clinical information only, which is worth clinical application.