Self-Supervised 3-D Semantic Representation Learning for Vision-and-Language Navigation.

In vision-and-language navigation (VLN) tasks, most current methods primarily utilize RGB images, overlooking the rich 3-D semantic data inherent to environments. To rectify this, we introduce a novel VLN framework that integrates 3-D semantic information into the navigation process. Our approach features a self-supervised training scheme that incorporates voxel-level 3-D semantic reconstruction to create a detailed 3-D semantic representation. A key component of this framework is a pretext task focused on region queries, which determines the presence of objects in specific 3-D areas. Following this, we devise an long short-term memory (LSTM)-based navigation model that is trained using our 3-D semantic representations. To maximize the utility of these 3-D semantic representations, we implement a cross-modal distillation strategy. This strategy encourages the RGB model's outputs to emulate those from the 3-D semantic feature network, enabling the concurrent training of both branches to merge RGB and 3-D semantic data effectively. Comprehensive evaluations on both the R2R and R4R datasets reveal that our method significantly enhances performance in VLN tasks.

Digital-Twin-Assisted Skill Learning for 3C Assembly Tasks.

The utilization of robots in computer, communication, and consumer electronics (3C) assembly has the potential to significantly reduce labor costs and enhance assembly efficiency. However, many typical scenarios in 3C assembly, such as the assembly of flexible printed circuits (FPCs), involve complex manipulations with long-horizon steps and high-precision requirements that cannot be effectively accomplished through manual programming or conventional skill-learning methods. To address this challenge, this article proposes a learning-based framework for the acquisition of complex 3C assembly skills assisted by a multimodal digital-twin environment. First, we construct a fully equivalent digital-twin environment based on the real-world counterpart, equipped with visual, tactile force, and proprioception information, and then collect multimodal demonstration data using virtual reality (VR) devices. Next, we construct a skill knowledge base through multimodal skill parsing of demonstration data, resulting in primitive policy sequences for achieving 3C assembly tasks. Finally, we train primitive policies via a combination of curriculum learning, residual reinforcement learning, and domain randomization methods and transfer the learned skill from the digital-twin environment to the real-world environment. The experiments are conducted to verify the effectiveness of our proposed method.

An efficient approach toward production of near-zigzag single-chirality carbon nanotubes.

Synthesizing single-walled carbon nanotubes (SWCNTs) with a narrow chirality distribution is essential for obtaining pure chirality materials through postgrowth sorting techniques. Using carbon monoxide chemical vapor deposition, we devise a ruthenium (Ru) catalyst supported by silica for the bulk production of SWCNTs containing only a few (n, m) species. The result is attributed to the limited carbon dissociation on the supported Ru clusters, favoring the growth of only small-diameter SWCNTs at comparable growth rates. The resulting materials expedite high-purity single chirality separation using gel chromatography, leading to unprecedented yields of 3.5% for (9, 1) and 5.2% for (9, 2) nanotubes, which surpass those separated from HiPco SWCNTs by two orders of magnitude. This work sheds light on the large-quantity synthesis of SWCNTs with enriched species beyond near-armchair ones for their high-yield separation.

Large-Area Flexible Carbon Nanofilms with Synergistically Enhanced Transmittance and Conductivity Prepared by Reorganizing Single-Walled Carbon Nanotube Networks.

Large-area flexible transparent conductive films (TCFs) are highly desired for future electronic devices. Nanocarbon TCFs are one of the most promising candidates, but some of their properties are mutually restricted. Here, a novel carbon nanotube network reorganization (CNNR) strategy, that is, the facet-driven CNNR (FD-CNNR) technique, is presented to overcome this intractable contradiction. The FD-CNNR technique introduces an interaction between single-walled carbon nanotube (SWNT) and Cu─-O. Based on the unique FD-CNNR mechanism, large-area flexible reorganized carbon nanofilms (RNC-TCFs) are designed and fabricated with A3-size and even meter-length, including reorganized SWNT (RSWNT) films and graphene and RSWNT (G-RSWNT) hybrid films. Synergistic improvement in strength, transmittance, and conductivity of flexible RNC-TCFs is achieved. The G-RSWNT TCF shows sheet resistance as low as 69 Ω sq-1 at 86% transmittance, FOM value of 35, and Young's modulus of ≈45 MPa. The high strength enables RNC-TCFs to be freestanding on water and easily transferred to any target substrate without contamination. A4-size flexible smart window is fabricated, which manifests controllable dimming and fog removal. The FD-CNNR technique can be extended to large-area or even large-scale fabrication of TCFs and can provide new insights into the design of TCFs and other functional films.

Multimodal information bottleneck for deep reinforcement learning with multiple sensors.

Reinforcement learning has achieved promising results on robotic control tasks but struggles to leverage information effectively from multiple sensory modalities that differ in many characteristics. Recent works construct auxiliary losses based on reconstruction or mutual information to extract joint representations from multiple sensory inputs to improve the sample efficiency and performance of reinforcement learning algorithms. However, the representations learned by these methods could capture information irrelevant to learning a policy and may degrade the performance. We argue that compressing information in the learned joint representations about raw multimodal observations is helpful, and propose a multimodal information bottleneck model to learn task-relevant joint representations from egocentric images and proprioception. Our model compresses and retains the predictive information in multimodal observations for learning a compressed joint representation, which fuses complementary information from visual and proprioceptive feedback and meanwhile filters out task-irrelevant information in raw multimodal observations. We propose to minimize the upper bound of our multimodal information bottleneck objective for computationally tractable optimization. Experimental evaluations on several challenging locomotion tasks with egocentric images and proprioception show that our method achieves better sample efficiency and zero-shot robustness to unseen white noise than leading baselines. We also empirically demonstrate that leveraging information from egocentric images and proprioception is more helpful for learning policies on locomotion tasks than solely using one single modality.

Surfactant Micelle-Driven High-Efficiency and High-Resolution Length Separation of Carbon Nanotubes for Electronic Applications.

High-efficiency extraction of long single-wall carbon nanotubes (SWCNTs) with excellent optoelectronic properties from SWCNT solution is critical for enabling their application in high-performance optoelectronic devices. Here, a straightforward and high-efficiency method is reported for length separation of SWCNTs by modulating the concentrations of binary surfactants. The results demonstrate that long SWCNTs can spontaneously precipitate for binary-surfactant but not for single-surfactant systems. This effect is attributed to the formation of compound micelles by binary surfactants that squeeze the free space of long SWCNTs due to their large excluded volumes. With this technique, it can readily separate near-pure long (≥500 nm in length, 99% in content) and short (≤500 nm in length, 98% in content) SWCNTs with separation efficiencies of 26% and 64%, respectively, exhibiting markedly greater length resolution and separation efficiency than those of previously reported methods. Thin-film transistors fabricated from extracted semiconducting SWCNTs with lengths >500 nm exhibit significantly improved electrical properties, including a 10.5-fold on-state current and 14.7-fold mobility, compared with those with lengths <500 nm. The present length separation technique is perfectly compatible with various surfactant-based methods for structure separations of SWCNTs and is significant for fabrication of high-performance electronic and optoelectronic devices.

Moderating Effect of Sex on the Relationship Between Physical Activity and Quality of Life in Coronary Heart Disease Patients in China.

PURPOSE: The aim of this study was to investigate the moderating effect of sex on the relationship between physical activity (PA) and quality of life (QoL) in Chinese patients with coronary heart disease (CHD) not participating in cardiac rehabilitation. METHODS: Chinese patients with CHD (aged 18-80 yr) were selected 12 mo after discharge from three Hebei Province tertiary hospitals. The International Physical Activity Questionnaire was used to assess PA in metabolic equivalents of energy (METs) and the Chinese Questionnaire of Quality of Life in Patients With Cardiovascular Disease was used to assess QoL. Data were analyzed using Student's t test and the χ 2 test, multivariant and hierarchical regression analysis, and simple slope analysis. RESULTS: Among 1162 patients with CHD studied between July 1 and November 30, 2017, female patients reported poorer QoL and lower total METs in weekly PA compared with male patients. Walking ( ß= .297), moderate-intensity PA ( ß= .165), and vigorous-intensity PA ( ß= .076) positively predicted QoL. Hierarchical regression analysis showed that sex moderates the relationship between walking ( ß= .195) and moderate-intensity PA ( ß= .164) and QoL, but not between vigorous-intensity PA ( ß= -.127) and QoL. Simple slope analysis revealed the standardized coefficients of walking on QoL were 0.397 (female t  = 8.210) and 0.338 (male t = 10.142); the standardized coefficients of moderate-intensity PA on QoL were 0.346 (female, t  = 7.000) and 0.175 (male, t = 5.033). CONCLUSIONS: Sex moderated the relationship between PA and QoL among patients with CHD in China. There was a greater difference in QoL for female patients reporting higher time versus those with lower time for both walking and moderate-intensity PA than for male patients.

Goal-Conditioned Hierarchical Reinforcement Learning With High-Level Model Approximation.

Hierarchical reinforcement learning (HRL) exhibits remarkable potential in addressing large-scale and long-horizon complex tasks. However, a fundamental challenge, which arises from the inherently entangled nature of hierarchical policies, has not been understood well, consequently compromising the training stability and exploration efficiency of HRL. In this article, we propose a novel HRL algorithm, high-level model approximation (HLMA), presenting both theoretical foundations and practical implementations. In HLMA, a Planner constructs an innovative high-level dynamic model to predict the k -step transition of the Controller in a subtask. This allows for the estimation of the evolving performance of the Controller. At low level, we leverage the initial state of each subtask, transforming absolute states into relative deviations by a designed operator as Controller input. This approach facilitates the reuse of subtask domain knowledge, enhancing data efficiency. With this designed structure, we establish the local convergence of each component within HLMA and subsequently derive regret bounds to ensure global convergence. Abundant experiments conducted on complex locomotion and navigation tasks demonstrate that HLMA surpasses other state-of-the-art single-level RL and HRL algorithms in terms of sample efficiency and asymptotic performance. In addition, thorough ablation studies validate the effectiveness of each component of HLMA.

Nurses' preparedness, opinions, barriers, and facilitators in responding to intimate partner violence: A mixed-methods study.

INTRODUCTION: Intimate partner violence (IPV) is associated with multiple adverse health consequences. Nurses (including midwives) are well positioned to identify patients subjected to IPV, and provide care, support, and referrals. However, studies about nursing response to IPV are limited especially in low- and middle-income countries (LMICs). The study aimed to examine nurses' perceived preparedness and opinions toward IPV and to identify barriers and facilitators in responding to IPV. DESIGN: An explanatory sequential mixed-methods study was conducted by collecting quantitative data first and explaining the quantitative findings with qualitative data. METHODS: The study was conducted in two tertiary general hospitals in northeastern (Shenyang city) and southwestern (Chengdu city) China with 1500 and 1800 beds, respectively. A total of 1071 survey respondents (1039 female [97.0%]) and 43 interview participants (34 female [79.1%]) were included in the study. An online survey was administered from September 3 to 23, 2020, using two validated scales from the Physician Readiness to Manage Intimate Partner Violence Survey. In-depth, semistructured interviews were conducted from September 15 to December 23, 2020, guided by the Consolidated Framework for Implementation Research. RESULTS: The survey respondents largely agreed with feeling prepared to manage IPV, e.g., respond to discourses (544 [50.8%] of 1071) and report to police (704 [65.7%] of 1071). The findings of surveyed opinions (i.e., Response competencies; Routine practice; Actual activities; Professionals; Victims; Alcohol/drugs) were mixed and intertwined with social desirability bias. The quantitative and qualitative data were consistent, contradicted, and supplemented. Key qualitative findings were revealed that may explain the quantitative results, including lack of actual preparedness, absence of IPV-related education, training, or practice, and socially desirable responses (especially those pertaining to China's Anti-domestic Violence Law). Commonly reported barriers (e.g., patients' reluctance to disclose; time constraints) and facilitators (e.g., patients' strong need for help; female nurses' gender advantage), as well as previously unreported barriers (e.g., IPV may become a workplace taboo if there are healthcare professionals known as victims/perpetrators of IPV) and facilitators (e.g., nurses' responses can largely meet the first-line support requirements even without formal education or training on IPV) were identified. CONCLUSIONS: Nurses may play a unique and important role in responding to IPV in LMICs where recognition is limited, education and training are absent, policies are lacking, and resources are scarce. Our findings support World Health Organization recommendations for selective screening. CLINICAL RELEVANCE: The study highlights the great potential of nurses for IPV prevention and intervention especially in LMICs. The identified barriers and facilitators are important evidence for developing multifaceted interventions to address IPV in the health sector.

Rich Action-Semantic Consistent Knowledge for Early Action Prediction.

Early action prediction (EAP) aims to recognize human actions from a part of action execution in ongoing videos, which is an important task for many practical applications. Most prior works treat partial or full videos as a whole, ignoring rich action knowledge hidden in videos, i.e., semantic consistencies among different partial videos. In contrast, we partition original partial or full videos to form a new series of partial videos and mine the Action-Semantic Consistent Knowledge (ASCK) among these new partial videos evolving in arbitrary progress levels. Moreover, a novel Rich Action-semantic Consistent Knowledge network (RACK) under the teacher-student framework is proposed for EAP. Firstly, we use a two-stream pre-trained model to extract features of videos. Secondly, we treat the RGB or flow features of the partial videos as nodes and their action semantic consistencies as edges. Next, we build a bi-directional semantic graph for the teacher network and a single-directional semantic graph for the student network to model rich ASCK among partial videos. The MSE and MMD losses are incorporated as our distillation loss to enrich the ASCK of partial videos from the teacher to the student network. Finally, we obtain the final prediction by summering the logits of different subnetworks and applying a softmax layer. Extensive experiments and ablative studies have been conducted, demonstrating the effectiveness of modeling rich ASCK for EAP. With the proposed RACK, we have achieved state-of-the-art performance on three benchmarks. The code is available at https://github.com/lily2lab/RACK.git.

Joint Power Control and Resource Allocation with Rate Fairness Consideration for SWIPT-Based Cognitive Two-Way Relay Networks.

This paper investigates the power control and resource allocation problem in a simultaneously wireless information and power transfer (SWIPT)-based cognitive two-way relay network, in which two secondary users exchange information through a power splitting (PS) energy harvesting (EH) cognitive relay node underlay in a primary network. To enhance the secondary networks's transmission ability without detriment to the primary network, we formulate an optimization to maximize the minimum transmission rates of the cognitive users by jointly optimizing power allocation at the sources, the time allocation of transmission frames and power splitting at the relay, under the constraint that the transmission power of the cognitive network is set not to exceed the primary user interference threshold to ensure primary work performance. To efficiently solve this problem, a sub-optimal algorithm named the joint power control and resource allocation (JPCRA) scheme is proposed, in which we decouple the non-convex problem into convex problems and use alternative steps in the optimization algorithm to get final solutions. Numerical results reveal that the proposed scheme enhances transmission fairness and outperforms three traditional schemes.

Informative Data Selection With Uncertainty for Multimodal Object Detection.

Noise has always been nonnegligible trouble in object detection by creating confusion in model reasoning, thereby reducing the informativeness of the data. It can lead to inaccurate recognition due to the shift in the observed pattern, that requires a robust generalization of the models. To implement a general vision model, we need to develop deep learning models that can adaptively select valid information from multimodal data. This is mainly based on two reasons. Multimodal learning can break through the inherent defects of single-modal data, and adaptive information selection can reduce chaos in multimodal data. To tackle this problem, we propose a universal uncertainty-aware multimodal fusion model. It adopts a multipipeline loosely coupled architecture to combine the features and results from point clouds and images. To quantify the correlation in multimodal information, we model the uncertainty, as the inverse of data information, in different modalities and embed it in the bounding box generation. In this way, our model reduces the randomness in fusion and generates reliable output. Moreover, we conducted a completed investigation on the KITTI 2-D object detection dataset and its derived dirty data. Our fusion model is proven to resist severe noise interference like Gaussian, motion blur, and frost, with only slight degradation. The experiment results demonstrate the benefits of our adaptive fusion. Our analysis on the robustness of multimodal fusion will provide further insights for future research.

Knowledge-Based Embodied Question Answering.

In this paper, we propose a novel Knowledge-based Embodied Question Answering (K-EQA) task, in which the agent intelligently explores the environment to answer various questions with the knowledge. Different from explicitly specifying the target object in the question as existing EQA work, the agent can resort to external knowledge to understand more complicated question such as "Please tell me what are objects used to cut food in the room?", in which the agent must know the knowledge such as "knife is used for cutting food". To address this K-EQA problem, a novel framework based on neural program synthesis reasoning is proposed, where the joint reasoning of the external knowledge and 3D scene graph is performed to realize navigation and question answering. Especially, the 3D scene graph can provide the memory to store the visual information of visited scenes, which significantly improves the efficiency for the multi-turn question answering. Experimental results have demonstrated that the proposed framework is capable of answering more complicated and realistic questions in the embodied environment. The proposed method is also applicable to multi-agent scenarios.

Length-Dependent Enantioselectivity of Carbon Nanotubes by Gel Chromatography.

High-purity enantiomer separation of chiral single-wall carbon nanotubes (SWCNTs) remains a challenge compared with electrical type and chirality separations due to the limited selectivities for both chirality and handedness, which is important for an exploration of their properties and practical applications. Here, we performed length fractionation for enantiomer-purified SWCNTs and found a phenomenon in which the enantioselectivities were higher for longer nanotubes than for shorter nanotubes due to length-dependent interactions with the gel medium, which provided an effective strategy of controlling nanotube length for high-purity enantiomer separation. Furthermore, we employed a gentler pulsed ultrasonication instead of traditional vigorous ultrasonication for preparation of a low-defect long SWCNT dispersion and achieved the enantiomer separation of single-chirality (6,5) SWCNTs with an ultrahigh enantiomeric purity of up to 98%, which was determined by using the linear relationship between the normalized circular dichroism intensity and the enantiomeric purity. Compared with all results reported previously, the present enantiomeric purity was significantly higher and reached the highest level reported to date. Due to the ultrahigh selectivity in both chirality and handedness, the two obtained enantiomers exhibited perfect symmetry in their circular dichroism spectra, which offers standardization for characterizations and evaluations of SWCNT enantiomers.

Self-Supervised Learning by Estimating Twin Class Distribution.

We present Twist, a simple and theoretically explainable self-supervised representation learning method by classifying large-scale unlabeled datasets in an end-to-end way. We employ a siamese network terminated by a softmax operation to produce twin class distributions of two augmented images. Without supervision, we enforce the class distributions of different augmentations to be consistent. However, simply minimizing the divergence between augmentations will generate collapsed solutions, i.e., outputting the same class distribution for all images. In this case, little information about the input images is preserved. To solve this problem, we propose to maximize the mutual information between the input image and the output class predictions. Specifically, we minimize the entropy of the distribution for each sample to make the class prediction assertive, and maximize the entropy of the mean distribution to make the predictions of different samples diverse. In this way, Twist can naturally avoid the collapsed solutions without specific designs such as asymmetric network, stop-gradient operation, or momentum encoder. As a result, Twist outperforms previous state-of-the-art methods on a wide range of tasks. Specifically on the semi-supervised classification task, Twist achieves 61.2% top-1 accuracy with 1% ImageNet labels using a ResNet-50 as backbone, surpassing previous best results by an improvement of 6.2%. Codes and pre-trained models are available at https://github.com/bytedance/TWIST.

Preparing high-concentration individualized carbon nanotubes for industrial separation of multiple single-chirality species.

Industrial production of single-chirality carbon nanotubes is critical for their applications in high-speed and low-power nanoelectronic devices, but both their growth and separation have been major challenges. Here, we report a method for industrial separation of single-chirality carbon nanotubes from a variety of raw materials with gel chromatography by increasing the concentration of carbon nanotube solution. The high-concentration individualized carbon nanotube solution is prepared by ultrasonic dispersion followed by centrifugation and ultrasonic redispersion. With this technique, the concentration of the as-prepared individualized carbon nanotubes is increased from about 0.19 mg/mL to approximately 1 mg/mL, and the separation yield of multiple single-chirality species is increased by approximately six times to the milligram scale in one separation run with gel chromatography. When the dispersion technique is applied to an inexpensive hybrid of graphene and carbon nanotubes with a wide diameter range of 0.8-2.0 nm, and the separation yield of single-chirality species is increased by more than an order of magnitude to the sub-milligram scale. Moreover, with present separation technique, the environmental impact and cost of producing single-chirality species are greatly reduced. We anticipate that this method promotes industrial production and practical applications of single-chirality carbon nanotubes in carbon-based integration circuits.

Hierarchical data structures for flowchart.

Flowcharts have broad applications in the fields of software development, engineering design, and scientific experimentation. Current flowchart data structure is mainly based on the adjacency list, cross-linked list, and adjacency matrix of the graph structure. Such design originated from the fact that any two nodes could have a connection relationship. But flowcharts have clear regularities, and their nodes have a certain inflow or outflow relationship. When graph structures such as an adjacency table or an adjacency matrix are used to store a flowchart, there is a large room for optimization in terms of traversal time and storage complexities, as well as usage convenience. In this paper we propose two hierarchical data structures for flowchart design. In the proposed structures, a flowchart is composed of levels, layers, and numbered nodes. The nodes between layers are connected according to a certain set of systematic design rules. Compared with the traditional graph data structures, the proposed schemes significantly reduce the storage space, improve the traversal efficiency, and resolve the problem of nesting between sub-charts. Experimental data based on flowchart examples used in this paper show that, compared with adjacency list, the hierarchical table data structure reduces the traversal time by 50% while their storage spaces are similar; compared with adjacency matrix, the hierarchical matrix data structure reduces the traversal time by nearly 70% and saves the storage space by about 50%. The proposed structures could have broad applications in flowchart-based software development, such as low-code engineering for smart industrial manufacturing.

Chirality-dependent electrical transport properties of carbon nanotubes obtained by experimental measurement.

Establishing the relationship between the electrical transport properties of single-wall carbon nanotubes (SWCNTs) and their structures is critical for the design of high-performance SWCNT-based electronic and optoelectronic devices. Here, we systematically investigated the effect of the chiral structures of SWCNTs on their electrical transport properties by measuring the performance of thin-film transistors constructed by eleven distinct (n, m) single-chirality SWCNT films. The results show that, even for SWCNTs with the same diameters but different chiral angles, the difference in the on-state current or carrier mobility could reach an order of magnitude. Further analysis indicates that the electrical transport properties of SWCNTs have strong type and family dependence. With increasing chiral angle for the same-family SWCNTs, Type I SWCNTs exhibit increasing on-state current and mobility, while Type II SWCNTs show the reverse trend. The differences in the electrical properties of the same-family SWCNTs with different chiralities can be attributed to their different electronic band structures, which determine the contact barrier between electrodes and SWCNTs, intrinsic resistance and intertube contact resistance. Our present findings provide an important physical basis for performance optimization and application expansion of SWCNT-based devices.

Recovery of vanadium with melamine in acidic medium.

Many hydro-metallurgical methods are developed to recover vanadium, while ammonium salt precipitation possesses the final step and it has threatened the environment. The key point is to find a new compound to replace ammonium salts without reducing vanadium recovery efficiency. Some compounds with -NH2 function groups have attracted our attention as they have similar function groups with ammonium salts. In this paper, the adsorption of vanadium with melamine is conducted. The results show that high adsorption efficiency can be achieved in a short time and melamine displays great performance in the recovery of all concentrations of vanadium. Response surface methodology (RSM) is used to optimize the reaction conditions and order the parameters: reaction temperature > concentration of vanadium > dosage of melamine > reaction time. 99.63% vanadium is adsorbed under optimized conditions: n(melamine)/n(V) = 0.6, reaction time of 60 min, 10 g/L vanadium solution and reaction temperature of 60°C. The successful application of melamine in the recovery of vanadium provides a new way for the utilization of melamine and also a glorious future for -NH2 compounds in the recovery heavy metals.

Individualized prognosis stratification in muscle invasive bladder cancer: A pairwise TP53-derived transcriptome signature.

TP53 is the most frequently mutated gene in muscle invasive bladder cancer (MIBC) and there are two gene signatures regarding TP53 developed for MIBC prognosis. However, they are limited to immune genes only and unable to be used individually across platforms due to their quantitative manners. We used 827 gene expression profiles from seven MIBC cohorts with varied platforms to build a pairwise TP53-derived transcriptome signature, 13 gene pairs (13-GPs). Since the 13-GPs model is a single sample prognostic predictor, it can be applied individually in practice and is applicable to any gene-expression platforms without specific normalization requirements. Survival difference between high-risk and low-risk patients stratified by the 13-GPs test was statistically significant (HR range: 2.26-2.76, all P < .0001). Discovery and validation sets showed that the 13-GPs was an independent prognostic factor after adjusting other clinical features (HR range: 2.21-2.82, all P < .05). Moreover, it was a potential supplement to the consensus molecular classification of MIBC to further stratify the LumP subtype (patients with better prognoses). High- and low-risk patients by the 13-GPs model presented distinct immune microenvironment and DDR mutation rates, suggesting that it might have the potential for immunotherapy. Being a general approach to other cancer types, this study demonstrated how we integrated gene variants with pairwise gene panels to build a single sample prognostic test in translational oncology.