Transcriptomic and neuroimaging data integration enhances machine learning classification of schizophrenia

Background

Schizophrenia is a polygenic disorder associated with changes in brain structure and function. Integrating macroscale brain features with microscale genetic data may provide a more complete overview of the disease etiology and may serve as potential diagnostic markers for schizophrenia.

Objective

We aim to systematically evaluate the impact of multi-scale neuroimaging and transcriptomic data fusion in schizophrenia classification models.

Methods

We collected brain imaging data and blood RNA sequencing data from 43 patients with schizophrenia and 60 age- and gender-matched healthy controls, and we extracted multi-omics features of macroscale brain morphology, brain structural and functional connectivity, and gene transcription of schizophrenia risk genes. Multi-scale data fusion was performed using a machine learning integration framework, together with several conventional machine learning methods and neural networks for patient classification.

Results

We found that multi-omics data fusion in conventional machine learning models achieved the highest accuracy (AUC ~0.76-0.92) in contrast to the single-modality models, with AUC improvements of 8.88 to 22.64%. Similar findings were observed for the neural network, showing an increase of 16.57% for the multimodal classification model (accuracy 71.43%) compared to the single-modal average. In addition, we identified several brain regions in the left posterior cingulate and right frontal pole that made a major contribution to disease classification.

Conclusion

We provide empirical evidence for the increased accuracy achieved by imaging genetic data integration in schizophrenia classification. Multi-scale data fusion holds promise for enhancing diagnostic precision, facilitating early detection and personalizing treatment regimens in schizophrenia.

High-Performance Hydrogen Evolution Reaction Catalytic Electrodes by Liquid Joule-Heating Growth

Despite the great progress in the research of integrated catalytic electrodes for hydrogen evolution reaction, the efficient preparation of high-performance catalytic electrodes with high current density remains a challenging issue. In this work, a metal (Pt)-amorphous oxide (NiO) heterostructure catalyst is successfully in situ grown on nickel foam using liquid Joule-heating. Based on the superhydrophilic surface of the electrode and its superior mechanical and chemical stability, the catalytic electrode exhibits excellent catalytic performance in alkaline electrolytes with only 100 mV overpotential to achieve 5000 mA cm-2 current density and maintains a stable performance of 500 h under a fixed current density of 1000 mA cm-2 . Further verification of the practical application of the Pt@NiO-Ni electrode in the alkaline electrolyzer is conducted. The results show that the alkaline water electrolyzer with NiFe layered double hydroxide as the anode and Pt@NiO-Ni as the cathode exhibits superior performance than the previously reported electrolyzers, with a current density of 1 A cm-2 already achieved at 1.75 V, which is even comparable to some anion exchange membrane water electrolyzers. These experimental results illustrate the strong applicability of Pt@NiO-Ni electrode at industrial scale current densities.

An Adaptive Model Filtering Algorithm Based on Grubbs Test in Federated Learning

Federated learning has been popular for its ability to train centralized models while protecting clients' data privacy. However, federated learning is highly susceptible to poisoning attacks, which can result in a decrease in model performance or even make it unusable. Most existing defense methods against poisoning attacks cannot achieve a good trade-off between robustness and training efficiency, especially on non-IID data. Therefore, this paper proposes an adaptive model filtering algorithm based on the Grubbs test in federated learning (FedGaf), which can achieve great trade-offs between robustness and efficiency against poisoning attacks. To achieve a trade-off between system robustness and efficiency, multiple child adaptive model filtering algorithms have been designed. Meanwhile, a dynamic decision mechanism based on global model accuracy is proposed to reduce additional computational costs. Finally, a global model weighted aggregation method is incorporated, which improves the convergence speed of the model. Experimental results on both IID and non-IID data show that FedGaf outperforms other Byzantine-robust aggregation rules in defending against various attack methods.

Structural and electronic properties of hydrogenated GaBi and InBi honeycomb monolayers with point defects

First-principles calculations are carried out to systematically investigate the structural and electronic properties of point defects in hydrogenated GaBi and InBi monolayers, including vacancies, antisites and Stone-Wales (SW) defects. Our results imply that the perfect H2-Ga(In)Bi is a semiconductor with a bandgap of 0.241 eV (0.265 eV) at the Γ point. The system turns into a metal by introducing a Ga(In) vacancy, substituting a Bi with a Ga(In) atom or substituting an In with a Bi atom. Other defect configurations can tune the bandgap value in the range from 0.09 eV to 0.3 eV. In particular, the exchange of neighboring Ga(In) and Bi increases the bandgap, meanwhile the spin splitting effect is preserved. All SW defects decrease the bandgap. The lowest formation energy of defects occurs when substituting a Ga(In) with a Bi atom and the values of SW defects vary from 0.98 eV to 1.77 eV.