Search NASA⌕ Search

SEARCH · Search NASA

Results for “Graph Processing”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 757 records · Page 42
All sourcesNASA NTRSDOE OSTI
NASA NTRS2012

Wide Range SET Pulse Measurement

A method for measuring a wide range of SET pulses is demonstrated. Use of dynamic logic, faster than ordinary CMOS, allows capture of short pulses. A weighted binning of SET lengths allows measurement of a wide range of pulse lengths with compact circuitry. A pulse-length-conservative pulse combiner tree routes SETs from combinational logic to the measurement circuit, allowing SET measurements in circuits that cannot easily be arranged in long chains. The method is applied to add-multiplex combinational logic, and to an array of NFET routing switches, at .35 micron. Pulses are captured in a chain of Domino Logic AND gates. Propagation through the chain is frozen on the trailing edge by dropping low the second "enable" input to the AND gates. Capacitive loading is increased in the latter stages to create an approximately logarithmic weighted binning, so that a broad range of pulse lengths can be captured with a 10 stage capture chain. Simulations show pulses can be captured which are 1/5th the length of those typically captured with leading edge triggered latch methods, and less than the length of those captured with a trailing edge latch method. After capture, the pulse pattern is transferred to an SEU protected shift register for readout. 64 instances of each of two types of logic are used as targets. One is a full adder with a 4 to 1 mux on its inputs. The other is a 4 x 4 NFET routing matrix. The outputs are passed through buffered XNOR comparators to identify pulses, which are merged in a buffered not-nand (OR) tree designed to avoid pulse absorption as much as possible. The output from each of the two test circuits are input into separate pulse measurement circuits. Test inputs were provided so that the circuit could be bench tested and calibrated. A third SET measurement circuit with no inputs was used to judge the contribution from direct hits on the measurement circuit. Heavy ions were used with an LET range from 12 to 176. At LET of 21 and below, the very small number of SETs were not significantly higher in the test over the control circuits. At higher LET the test circuit SETs are one or two orders of magnitude greater than for the control circuit. The NFET circuit produces more and slightly longer SETs as expected. But the differences do not appear to be significant enough to modify strategies now used to avoid capture of SETs in chips such as FPGAs. Complete data and graphs will be in the full paper / presentation. In the summary figure below left, NOCL is the reference circuit without any input, and number of stages triggered is plotted. Simulation at right shows the smallest pulse captured (stage 2) at about 300 ps. Our conclusion is that the method is promising, but that improvements in the merge network are desirable before applying in a deep submicron process

Shuler, Robert L.↗
NASA NTRSDate not supplied

Biological Data for Deep Space Mission Support

Increased biomedical risks and challenges associated with deep space missions (cis-Lunar, Mars transit, Mars surface) require new knowledge discovery and development of novel ecosystem and biomedical support capabilities. This paradigm shift supporting distant and long-duration missions requires biological data to be findable, accessible, interoperable, reusable (FAIR), and maximally open-access (i.e., there is a data governance continuum from closed to mediated to embargoed to open). The NASA “Open Science Data Repositories” (OSDR) aims to meet scientific, technical, and operational spaceflight needs, and offers the ability to upload, download, search, share, analyze, and visualize data across physiological, behavioral, ‘omics, and environmental monitoring telemetry datasets. OSDR includes NASA GeneLab, NASA Ames Life Sciences Data Archive (ALSDA), and NASA Biological Institutional Scientific Collection (NBISC). In the past year, ALSDA has undergone a transformation in its data collection, curation, and architecture methods. Standardizing non-genomic (phenotypic) datasets was, and will continue to be, a challenge because of their diverse nature (e.g., molecular, cellular, tissue, whole organism behavior; micro-computed tomography, intraocular pressure, fluorescence microscopy, western blot, ultrasonography; tabular, images, video). This year ALSDA, alongside GeneLab, introduced the Biological Data Management Environment (BDME) with the purpose to accept submission of data from space relevant experiments including spaceflight, radiation, simulated gravity, gravitropism, isolation and confinement, hostile closed environments and/or distance from Earth. In addition to bringing together omics, phenotypic, physiological, bioimaging, and behavioral data into one repository. By integrating with GeneLab a multi-project submission portal aims to reduce the burden on PIs submitting data and enabling the discovery of both omics and phenotypic data. The purpose of ALSDA is to collect, curate, and make all non-human space-relevant biological data maximally findable, accessible, interoperable, and reusable (FAIR). These scope of ALSDA data collected and submitted by PIs include study design metadata, subject metadata, assay metadata (parameters), raw and processed assay data, assay imagery/video, and subject-experienced mission data telemetry (radiation, temperature, humidity, acoustics, vibrations, etc.). In 2021, a community of researchers rallied to form the ALSDA Analysis Working Group (AWG) and provided scientific consensus on dataset sample and assay metadata. The community and excitement around the ALSDA/OSDR system has already led to several data reuse studies, demonstrating value using machine learning (ML), knowledge graphs, and meta-analysis approaches.

space biology↗
NASA NTRSDate not supplied

Biological Data for Deep Space Mission Support

Increased biomedical risks and challenges associated with deep space missions (cis-Lunar, Mars transit, Mars surface) require new knowledge discovery and development of novel ecosystem and biomedical support capabilities. This paradigm shift supporting distant and long-duration missions requires biological data to be findable, accessible, interoperable, reusable (FAIR), and maximally open-access (i.e., there is a data governance continuum from closed to mediated to embargoed to open). The NASA “Open Science Data Repositories” (OSDR) aims to meet scientific, technical, and operational spaceflight needs, and offers the ability to upload, download, search, share, analyze, and visualize data across physiological, behavioral, ‘omics, and environmental monitoring telemetry datasets. OSDR includes NASA GeneLab, NASA Ames Life Sciences Data Archive (ALSDA), and NASA Biological Institutional Scientific Collection (NBISC). In the past year, ALSDA has undergone a transformation in its data collection, curation, and architecture methods. Standardizing non-genomic (phenotypic) datasets was, and will continue to be, a challenge because of their diverse nature (e.g., molecular, cellular, tissue, whole organism, behavior; micro-computed tomography, intraocular pressure, fluorescence microscopy, western blot, ultrasonography; tabular, images, video). This year ALSDA, alongside GeneLab, introduced the Biological Data Management Environment (BDME) with the purpose to accept submission of data from space relevant experiments including spaceflight, radiation, simulated gravity, gravitropism, isolation and confinement, hostile closed environments and/or distance from Earth. In addition to bringing together omics, phenotypic, physiological, bioimaging, and behavioral data into one repository. By integrating with GeneLab a multi-project submission portal aims to reduce the burden on PIs submitting data and enabling the discovery of both omics and phenotypic data. The purpose of ALSDA is to collect, curate, and make all non-human space-relevant biological data maximally findable, accessible, interoperable, and reusable (FAIR). These scope of ALSDA data collected and submitted by PIs include study design metadata, subject metadata, assay metadata (parameters), raw and processed assay data, assay imagery/video, and subject-experienced mission data telemetry (radiation, temperature, humidity, acoustics, vibrations, etc.). In 2021, a community of researchers rallied to form the ALSDA Analysis Working Group (AWG) and provided scientific consensus on dataset sample and assay metadata. The community and excitement around the ALSDA/OSDR system has already led to several data reuse studies, demonstrating value using machine learning (ML), knowledge graphs, and meta-analysis approaches.

space biology↗
DOE OSTI2018

Deep Learning for Scientific Discovery

Deep learning has positively impacted fields were perceptive tasks are paramount, such as computer vision and natural language processing. Tremendous strides have also been made in deep reinforcement learning towards robotics and gaming. Given the success in these fields, it is apparent that deep learning may also accelerate scientific discovery. Herein, we discuss several topic areas where modern representation learning is driving innovation. This article presents a) learning non-linear models from data, b) learning scientific laws from models, c) deep learning on graphs, and d) approaches to scalable deep learning.

deep learning, scientific computing, koopman, mate↗
DOE OSTI2026

Estimating CO 2 fluxes through integrating spatial and temporal input layers via deep learning algorithms

Background Accurate estimation of net ecosystem exchange of CO 2 fluxes (Fc) is essential for understanding carbon cycle processes and assessing ecosystem carbon budgets. However, conventional modeling approaches often emphasize temporal dynamics while overlooking the pronounced spatial heterogeneity within the footprint of eddy covariance (EC) towers, potentially limiting predictive accuracy and interpretability of Fc estimates. To address this challenge, we developed a spatiotemporal model that integrates high-resolution footprint-weighted spatial information with sequential environmental drivers. Results The integrated model combines a deeper graph convolutional network to characterize fine-scale spatial variability within EC footprints and a gated recurrent unit network to capture temporal dependencies in biophysical conditions. Using multi-year flux tower observations, remote sensing vegetation indices and footprint modeling, we evaluate the proposed method across three land cover types. This spatiotemporal model consistently outperforms temporal-only and spatial-only baselines, achieving the highest overall accuracy (R 2 = 0.9569) and the lowest RMSE (1.8128 μmol m −2 s −1 ) and MAE (1.1939 μmol m −2 s −1 ). Performance gains are particularly evident in ecosystems with strong vegetation heterogeneity, where spatial structure substantially modulates Fc variability. Conclusions This study demonstrates the importance of joint modeling spatial heterogeneity and temporal dynamics for improving Fc estimation and provides a robust method for advancing footprint-based Fc estimates across diverse ecosystems, supporting refined assessments of terrestrial carbon fluxes, and enhancing scientific foundations for carbon studies.

CO2 flux estimate↗
NASA NTRS2016

Evaluation of the Effect of Surface Finish on High-Cycle Fatigue of SLM-IN718

The surface finish of parts produced by additive manufacturing processes is much rougher than the surface finish generated by machining processes, and a rougher surface can reduce the fatigue strength of a part. This paper discusses an effort to quantify that reduction of strength in high-cycle fatigue for selective laser melt (SLM) coupons. A high-cycle fatigue (HCF) knockdown factor was estimated for Inconel 718, manufactured with the SLM process. This factor is the percentage reduction from the maximum stress in fatigue for low-stress ground (LSG) specimens to the maximum stress of those left with the original surface condition at the same fatigue life. Specimens were provided by a number of vendors, free to use their "best practice"; only one heat treat condition was considered; and several test temperatures were characterized, including room temperature, 800F, 1000F, and 1200F. The 1000F data had a large variance, and was omitted from consideration in this document. A first method used linear approximations extracted from the graphs, and only where data was available for both. A recommended knockdown factor of the as-built surface condition (average roughness of approximately 245 micro-inches/inch) versus low-stress ground condition (roughness no more than 4 micro-inches/inch) was established at approximately 1/3 or 33%. This is to say that for the as-built surface condition, a maximum stress of 2/3 of the stress for LSG can be expected to produce a similar life in the as-built surface condition. In this first evaluation, the knockdown factor did not appear to be a function of temperature. A second approach, the "KP method", incorporated the surface finish measure into a new parameter termed the pseudo-stress intensity factor, Kp, which was formulated to be similar to the fracture mechanics stress intensity factor. Using Kp, the variance seemed to be reduced across all sources, and knockdown factors were estimated using Kp over the range where data occurred. A plot of the results suggests that the knockdown factor is a function of temperature, and that for low lives the knockdown might be lower than the knockdown observed above about one million cycles, where it tended to stabilize. This was not universal for all temperatures tested. The higher temperature tests are thought to be influenced by the test temperature, which perhaps continued the aging process. Further evaluation of the method is suggested.

Lambert, D. M.↗
DOE OSTI2025

yProv4ML: Effortless provenance tracking for machine learning systems

The rapid growth in interest in deep learning and foundation models (FMs) in particular, has attracted the attention of a diverse range of researchers thanks to their generalization ability. However, the advent of these techniques has also brought to light the lack of transparency and rigor in the way development is pursued. In particular, the inability to determine the number of epochs and other hyperparameters in advance presents challenges in identifying the best model. To address this challenge, machine learning frameworks such as MLFlow can automate the collection of this type of information. However, these tools capture data using proprietary formats and pose little attention to lineage. This paper proposes yProv4ML, a framework that captures provenance information generated during machine learning processes in PROV-JSON format, with minimal code modification.

Machine learning↗
NASA NTRS1992

Mentat/A: Medium grain parallel processing

The objective of this project is to test the Algorithm to Architecture Mapping Model (ATAMM) firing rules using the Mentat run-time system and the Mentat Programming Language (MPL). A special version of Mentat, Mentat/A (Mentat/ATAMM) was constructed. This required changes to: (1) modify the run-time system to control queue length and inhibit actor firing until required data tokens are available and space is available in the input queues of all of the direct descendent actors; (2) disallow the specification of persistent object classes in the MPL; and (3) permit only decision free graphs in the MPL. We were successful in implementing the spirit of the plan, although some goals changed as we came to better understand the problem. We report on what we accomplished and the lessons we learned. The Mentat/A run-time system is discussed, and we briefly present the compiler. We present results for three applications and conclude with a summary and some observations. Appendix A contains a list of technical reports and published papers partially supported by the grant. Appendix B contains listings for the three applications.

Grimshaw, Andrew S.↗
NASA NTRS1987

The effect of the isotopic composition of oxygen on the non-mass-dependent isotopic fractionation in the formation of ozone by discharge of O2

Isotopic fractionation processes in the formation of O3 by 20-kV 60-kHz discharge of O2 in a fused silica container at 77 K are investigated experimentally, with a focus on the effects of changes in the isotopic composition of O2, the O2 pressure, the degree of O3 formation, and the discharge geometry on the isotope abundances in the postdischarge O2 and O3. The results are presented in tables and graphs and compared with the predictions of theoretical models involving vibrational anharmonicity and symmetry effects on O3 predissociation. In experiments using O2 enriched with O-17 and O-18, the heavy isotope is enriched in the residual O2 and not in the O3, whereas the opposite is true when atmospheric isotope abundances are used. It is inferred that the predissociation rate depends on the symmetry and mass of the O3 molecules and not on self-shielding or symmetry effects in the reactant O2. The potential value of the present findings for studies of anomalous isotope abundances in meteorites is indicated.

Yang, Jongmann↗
NASA NTRS1987

The effect of pressure and excitation energy on the isotopic fractionation in the formation of ozone by discharge of O2

Isotopic fractionation processes in the formation of O3 by discharge (at 20 kV and 3 MHz) of O2 in a fused silica container at 77 K are investigated experimentally, with a focus on the effects of changes in the isotopic abundances in the predischarge O2, the O2 pressure, and the excitation energy on the isotope abundances in the postdischarge O2 and O3. The results are presented in extensive tables and graphs and characterized in detail, and good general agreement is found with data obtained using 60-kHz excitation (Yang and Epstein, 1987). It is shown that O-17 and O-18 enrichment of the residual O2 (when the predischarge gas is enriched) or of the O3 (when atmospheric isotope abundances are used) can occur independently of one another. It is inferred that the rate of anharmonic predissociation of vibrationally hot O3 depends on the symmetry and mass of the O3 molecules. The significance of the present findings for studies of anomalous oxygen isotope abundances in meteorites is indicated.

Yang, Jongmann↗
NASA NTRS2000

Analysis of Phase-Type Stochastic Petri Nets With Discrete and Continuous Timing

The Petri net formalism is useful in studying many discrete-state, discrete-event systems exhibiting concurrency, synchronization, and other complex behavior. As a bipartite graph, the net can conveniently capture salient aspects of the system. As a mathematical tool, the net can specify an analyzable state space. Indeed, one can reason about certain qualitative properties (from state occupancies) and how they arise (the sequence of events leading there). By introducing deterministic or random delays, the model is forced to sojourn in states some amount of time, giving rise to an underlying stochastic process, one that can be specified in a compact way and capable of providing quantitative, probabilistic measures. We formalize a new non-Markovian extension to the Petri net that captures both discrete and continuous timing in the same model. The approach affords efficient, stationary analysis in most cases and efficient transient analysis under certain restrictions. Moreover, this new formalism has the added benefit in modeling fidelity stemming from the simultaneous capture of discrete- and continuous-time events (as opposed to capturing only one and approximating the other). We show how the underlying stochastic process, which is non-Markovian, can be resolved into simpler Markovian problems that enjoy efficient solutions. Solution algorithms are provided that can be easily programmed.

Jones, Robert L.↗
NASA NTRS2011

Antenna Near-Field Probe Station Scanner

A miniaturized antenna system is characterized non-destructively through the use of a scanner that measures its near-field radiated power performance. When taking measurements, the scanner can be moved linearly along the x, y and z axis, as well as rotationally relative to the antenna. The data obtained from the characterization are processed to determine the far-field properties of the system and to optimize the system. Each antenna is excited using a probe station system while a scanning probe scans the space above the antenna to measure the near field signals. Upon completion of the scan, the near-field patterns are transformed into far-field patterns. Along with taking data, this system also allows for extensive graphing and analysis of both the near-field and far-field data. The details of the probe station as well as the procedures for setting up a test, conducting a test, and analyzing the resulting data are also described.

Zaman, Afroz J.↗
DOE OSTI2024

Geometry-complete diffusion for 3D molecule generation and optimization

Abstract Generative deep learning methods have recently been proposed for generating 3D molecules using equivariant graph neural networks (GNNs) within a denoising diffusion framework. However, such methods are unable to learn important geometric properties of 3D molecules, as they adopt molecule-agnostic and non-geometric GNNs as their 3D graph denoising networks, which notably hinders their ability to generate valid large 3D molecules. In this work, we address these gaps by introducing the Geometry-Complete Diffusion Model (GCDM) for 3D molecule generation, which outperforms existing 3D molecular diffusion models by significant margins across conditional and unconditional settings for the QM9 dataset and the larger GEOM-Drugs dataset, respectively. Importantly, we demonstrate that GCDM’s generative denoising process enables the model to generate a significant proportion of valid and energetically-stable large molecules at the scale of GEOM-Drugs, whereas previous methods fail to do so with the features they learn. Additionally, we show that extensions of GCDM can not only effectively design 3D molecules for specific protein pockets but can be repurposed to consistently optimize the geometry and chemical composition of existing 3D molecules for molecular stability and property specificity, demonstrating new versatility of molecular diffusion models. Code and data are freely available on GitHub .

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2024

Deeplynx Dag Repository

The DeepLynx DAG repository will contain several Airflow DAGs (Directed Acyclic Graphs) which will be used in the context of DeepLynx's deployed Apache Airflow instance. These DAGs will be used for multiple data management tasks for DeepLynx data, including but not limited to: - bringing data from various sources and tools into DeepLynx - managing sequential data workflows, such as running Python scripts on data to perform analysis and returning the results to DeepLynx - performing any necessary transformation or pre-processing on data coming into DeepLynx from external sources or out of DeepLynx to go to external applications

Brownlee, JarenM.↗
DOE OSTI2026

Intelligent Experiments Through Real-time AI: Fast Data Processing and Autonomous Detector Control for sPHENIX and Future EIC Detectors (Final Report)

The overall vision of this project was to integrate real-time artificial intelligence (AI) directly into the data acquisition and detector-control systems of nuclear physics experiments, including both fast online event selection and an autonomous detector-control feedback loop. The work carried out under the award focused on the fast online event-selection half of that vision: the efficient recording of low-momentum heavy-flavor (HF) hadron decays in proton-proton collisions at the sPHENIX experiment at the Relativistic Heavy Ion Collider (RHIC)—an observable that requires fast tracking and topological trigger selection not previously demonstrated at RHIC, and that is essential for QCD studies at future facilities such as the Electron-Ion Collider (EIC). The autonomous detector-control (GPU-based feedback) component named in the project title remained a design concept and was not implemented under this award. The Massachusetts Institute of Technology (MIT) group led the offline simulation and data processing needed to train the machine-learning (ML) models, the translation of trained models to Field-Programmable Gate Array (FPGA) firmware using the hls4ml framework, and the physics validation of heavy-flavor reconstruction. Over the award period, the team developed and hardware-tested the principal components of an AI-based heavy-flavor trigger on simulated and recorded sPHENIX tracker data: a software Bipartite Graph Attention Network (BiGAT) trigger model reaching > 95% signal efficiency at 99% background rejection; an FPGA-native hit clusterizer matching the offline clustering; smaller networks synthesized to FPGA within the required sub-10 µs latency; and an assembled decoder–clusterizer–inference firmware chain exercised on the FELIX readout board. A complete, fully integrated hardware demonstrator was not finished within the award period. This report documents the project goals, the MIT group’s contributions, the technical accomplishments, and the outlook toward applications at the future EIC ePIC detector.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
NASA NTRS2000

A Parallel Genetic Algorithm for Automated Electronic Circuit Design

Parallelized versions of genetic algorithms (GAs) are popular primarily for three reasons: the GA is an inherently parallel algorithm, typical GA applications are very compute intensive, and powerful computing platforms, especially Beowulf-style computing clusters, are becoming more affordable and easier to implement. In addition, the low communication bandwidth required allows the use of inexpensive networking hardware such as standard office ethernet. In this paper we describe a parallel GA and its use in automated high-level circuit design. Genetic algorithms are a type of trial-and-error search technique that are guided by principles of Darwinian evolution. Just as the genetic material of two living organisms can intermix to produce offspring that are better adapted to their environment, GAs expose genetic material, frequently strings of 1s and Os, to the forces of artificial evolution: selection, mutation, recombination, etc. GAs start with a pool of randomly-generated candidate solutions which are then tested and scored with respect to their utility. Solutions are then bred by probabilistically selecting high quality parents and recombining their genetic representations to produce offspring solutions. Offspring are typically subjected to a small amount of random mutation. After a pool of offspring is produced, this process iterates until a satisfactory solution is found or an iteration limit is reached. Genetic algorithms have been applied to a wide variety of problems in many fields, including chemistry, biology, and many engineering disciplines. There are many styles of parallelism used in implementing parallel GAs. One such method is called the master-slave or processor farm approach. In this technique, slave nodes are used solely to compute fitness evaluations (the most time consuming part). The master processor collects fitness scores from the nodes and performs the genetic operators (selection, reproduction, variation, etc.). Because of dependency issues in the GA, it is possible to have idle processors. However, as long as the load at each processing node is similar, the processors are kept busy nearly all of the time. In applying GAs to circuit design, a suitable genetic representation 'is that of a circuit-construction program. We discuss one such circuit-construction programming language and show how evolution can generate useful analog circuit designs. This language has the desirable property that virtually all sets of combinations of primitives result in valid circuit graphs. Our system allows circuit size (number of devices), circuit topology, and device values to be evolved. Using a parallel genetic algorithm and circuit simulation software, we present experimental results as applied to three analog filter and two amplifier design tasks. For example, a figure shows an 85 dB amplifier design evolved by our system, and another figure shows the performance of that circuit (gain and frequency response). In all tasks, our system is able to generate circuits that achieve the target specifications.

Long, Jason D.↗
DOE OSTI2022

Physics-informed graph neural networks for predicting cetane number with systematic data quality analysis

Designing alternative fuels for advanced compression ignition engines necessitates a predictive model for cetane number (CN). In this study, the physics-informed graph neural networks are introduced for a reliable CN prediction by considering molecular features pertinent to the physical properties of molecules that affect CN. The reliability of measured data is another key factor to consider for improving the predictive model. Various experimental instruments for measuring CN exist, including standard and non-standard methods. In this regard, a systematic data quality analysis was carried out for the total 630 CNs collected from literature and new measurements in this study using Advanced Fuel Ignition Delay Analyzer (AFIDA). The results from this data curation process were reflected in the model by imposing lower sample weights on the data coming from less reliable measurement techniques. This approach effectively maximized the prediction accuracy while incorporating data from all available sources. Using the sample weights decreased the mean absolute error (MAE) up to 0.8 CN units. The accuracy was also improved by introducing the CN-related physical properties (the number of hydrogen bond donors and acceptors); the test set MAE is 5.74 and 7.01 for the model with and without such properties, respectively. Investigating molecular structural effects on CN was also carried out to gain chemical insights into factors used to design new fuel candidates. The dimensionality reduction analysis of feature vectors showed a clear clustering in terms of functional groups and CN and the structural effect derived from the model was consistent with the physicochemical insights. Finally, this physics-informed model and data curation would be helpful for accurate CN prediction and inform rational fuel design.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Discriminative analysis of schizophrenia patients using graph convolutional networks: A combined multimodal MRI and connectomics analysis

Introduction Recent studies in human brain connectomics with multimodal magnetic resonance imaging (MRI) data have widely reported abnormalities in brain structure, function and connectivity associated with schizophrenia (SZ). However, most previous discriminative studies of SZ patients were based on MRI features of brain regions, ignoring the complex relationships within brain networks. Methods We applied a graph convolutional network (GCN) to discriminating SZ patients using the features of brain region and connectivity derived from a combined multimodal MRI and connectomics analysis. Structural magnetic resonance imaging (sMRI) and resting-state functional magnetic resonance imaging (rs-fMRI) data were acquired from 140 SZ patients and 205 normal controls. Eighteen types of brain graphs were constructed for each subject using 3 types of node features, 3 types of edge features, and 2 brain atlases. We investigated the performance of 18 brain graphs and used the TopK pooling layers to highlight salient brain regions (nodes in the graph). Results The GCN model, which used functional connectivity as edge features and multimodal features (sMRI + fMRI) of brain regions as node features, obtained the highest average accuracy of 95.8%, and outperformed other existing classification studies in SZ patients. In the explainability analysis, we reported that the top 10 salient brain regions, predominantly distributed in the prefrontal and occipital cortices, were mainly involved in the systems of emotion and visual processing. Discussion Our findings demonstrated that GCN with a combined multimodal MRI and connectomics analysis can effectively improve the classification of SZ at an individual level, indicating a promising direction for the diagnosis of SZ patients. The code is available at https://github.com/CXY-scut/GCN-SZ.git .

Chen, Xiaoyi↗