Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

General General

Predicting the Risk Genes of Autism Spectrum Disorders.

In Frontiers in genetics ; h5-index 62.0

Autism spectrum disorder (ASD) refers to a wide spectrum of neurodevelopmental disorders that emerge during infancy and continue throughout a lifespan. Although substantial efforts have been made to develop therapeutic approaches, core symptoms persist lifelong in ASD patients. Identifying the brain temporospatial regions where the risk genes are expressed in ASD patients may help to improve the therapeutic strategies. Accordingly, this work aims to predict the risk genes of ASD and identify the temporospatial regions of the brain structures at different developmental time points for exploring the specificity of ASD gene expression in the brain that would help in possible ASD detection in the future. A dataset consisting of 13 developmental stages ranging from 8 weeks post-conception to 8 years from 26 brain structures was retrieved from the BrainSpan atlas. This work proposes a support vector machine-based risk gene prediction method ASD-Risk to distinguish the risk genes of ASD and non-ASD genes. ASD-Risk used an optimal feature selection algorithm called inheritable bi-objective combinatorial genetic algorithm to identify the brain temporospatial regions for prediction of the risk genes of ASD. ASD-Risk achieved a 10-fold cross-validation accuracy, sensitivity, specificity, area under a receiver operating characteristic curve, and a test accuracy of 81.83%, 0.84, 0.79, 0.84, and 72.27%, respectively. We prioritized the temporospatial features according to their contribution to the prediction accuracy. The top identified temporospatial regions of the brain for risk gene prediction included the posteroventral parietal cortex at 13 post-conception weeks feature. The identified temporospatial features would help to explore the risk genes that are specifically expressed in different brain regions of ASD patients.

Lin Yenching, Yerukala Sathipati Srinivasulu, Ho Shinn-Ying


autism spectrum disorders, feature selection, gene expression profiles, machine learning, risk gene prediction

Pathology Pathology

A Machine Learning Method to Identify Genetic Variants Potentially Associated With Alzheimer's Disease.

In Frontiers in genetics ; h5-index 62.0

There is hope that genomic information will assist prediction, treatment, and understanding of Alzheimer's disease (AD). Here, using exome data from ∼10,000 individuals, we explore machine learning neural network (NN) methods to estimate the impact of SNPs (i.e., genetic variants) on AD risk. We develop an NN-based method (netSNP) that identifies hundreds of novel potentially protective or at-risk AD-associated SNPs (along with an effect measure); the majority with frequency under 0.01. For case individuals, the number of "protective" (or "at-risk") netSNP-identified SNPs in their genome correlates positively (or inversely) with their age of AD diagnosis and inversely (or positively) with autopsy neuropathology. The effect measure increases correlations. Simulations suggest our results are not due to genetic linkage, overfitting, or bias introduced by netSNP. These findings suggest that netSNP can identify SNPs associated with AD pathophysiology that may assist with the diagnosis and mechanistic understanding of the disease.

Monk Bradley, Rajkovic Andrei, Petrus Semar, Rajkovic Aleks, Gaasterland Terry, Malinow Roberto


Alzheimer’s, disease, machine learning, neural network, polygenic

General General

Temporal Integration of Text Transcripts and Acoustic Features for Alzheimer's Diagnosis Based on Spontaneous Speech.

In Frontiers in aging neuroscience ; h5-index 64.0

Background: Advances in machine learning (ML) technology have opened new avenues for detection and monitoring of cognitive decline. In this study, a multimodal approach to Alzheimer's dementia detection based on the patient's spontaneous speech is presented. This approach was tested on a standard, publicly available Alzheimer's speech dataset for comparability. The data comprise voice samples from 156 participants (1:1 ratio of Alzheimer's to control), matched by age and gender. Materials and Methods: A recently developed Active Data Representation (ADR) technique for voice processing was employed as a framework for fusion of acoustic and textual features at sentence and word level. Temporal aspects of textual features were investigated in conjunction with acoustic features in order to shed light on the temporal interplay between paralinguistic (acoustic) and linguistic (textual) aspects of Alzheimer's speech. Combinations between several configurations of ADR features and more traditional bag-of-n-grams approaches were used in an ensemble of classifiers built and evaluated on a standardised dataset containing recorded speech of scene descriptions and textual transcripts. Results: Employing only semantic bag-of-n-grams features, an accuracy of 89.58% was achieved in distinguishing between Alzheimer's patients and healthy controls. Adding temporal and structural information by combining bag-of-n-grams features with ADR audio/textual features, the accuracy could be improved to 91.67% on the test set. An accuracy of 93.75% was achieved through late fusion of the three best feature configurations, which corresponds to a 4.7% improvement over the best result reported in the literature for this dataset. Conclusion: The proposed combination of ADR audio and textual features is capable of successfully modelling temporal aspects of the data. The machine learning approach toward dementia detection achieves best performance when ADR features are combined with strong semantic bag-of-n-grams features. This combination leads to state-of-the-art performance on the AD classification task.

Martinc Matej, Haider Fasih, Pollak Senja, Luz Saturnino


“Alzheimers dementia detection”, acoustic features, language, lexical features, machine learning, natural language processing, speech, speech processing

General General

Reaction-Diffusion Model-Based Research on Formation Mechanism of Neuron Dendritic Spine Patterns.

In Frontiers in neurorobotics

The pattern abnormalities of dendritic spine, tiny protrusions on neuron dendrites, have been found related to multiple nervous system diseases, such as Parkinson's disease and schizophrenia. The determination of the factors affecting spine patterns is of vital importance to explore the pathogenesis of these diseases, and further, search the treatment method for them. Although the study of dendritic spines is a hot topic in neuroscience in recent years, there is still a lack of systematic study on the formation mechanism of its pattern. This paper provided a reinterpretation of reaction-diffusion model to simulate the formation process of dendritic spine, and further, study the factors affecting spine patterns. First, all four classic shapes of spines, mushroom-type, stubby-type, thin-type, and branched-type were reproduced using the model. We found that the consumption rate of substrates by the cytoskeleton is a key factor to regulate spine shape. Moreover, we found that the density of spines can be regulated by the amount of an exogenous activator and inhibitor, which is in accordance with the anatomical results found in hippocampal CA1 in SD rats with glioma. Further, we analyzed the inner mechanism of the above model parameters regulating the dendritic spine pattern through Turing instability analysis and drew a conclusion that an exogenous inhibitor and activator changes Turing wavelength through which to regulate spine densities. Finally, we discussed the deep regulation mechanisms of several reported regulators of dendritic spine shape and densities based on our simulation results. Our work might evoke attention to the mathematic model-based pathogenesis research for neuron diseases which are related to the dendritic spine pattern abnormalities and spark inspiration in the treatment research for these diseases.

Jia Yiqing, Zhao Qili, Yin Hongqiang, Guo Shan, Sun Mingzhu, Yang Zhuo, Zhao Xin


Turing instability, branching morphogenesis, dendritic spine, glioma, reaction-diffusion model

General General

Industrial application of QbD and NIR chemometric models in quality improvement of immediate release tablets.

In Saudi pharmaceutical journal : SPJ : the official publication of the Saudi Pharmaceutical Society

Quality by Design (QbD) and chemometric models are different sides of the same coin. While QbD models utilize experimentally designed settings for optimization of some quality attributes, these settings can also be utilized for chemometric prediction of the same attributes. We aimed to synchronize optimization of comparative dissolution results of carvedilol immediate release tablets with chemometric prediction of dissolution profile and content uniformity of the product. As an industrial application, selection of variables for optimization was done by performing risk assessment utilizing the archived product records at the pharmaceutical site. Experimental tablets were produced with 20 different settings with the variables being contents of sucrose, sodium starch glycolate, lactose monohydrate, and avicel Ph 101. Contents of the excipients were modelled with F1 dissimilarity factor and F2 similarity factor in HCL, acetate, and USP dissolution media to determine the design space. We initiatively utilized Partial Least Square based Structural Equation Modelling (PLS-SEM) to explore how the excipients and their NIR records explained dissolution of the product. Finally, the optimized formula was utilized with varied content of carvedilol for chemometric prediction of the content uniformity.

Ramadan Ahmed, Basalious Emad B, Abdallah Mohamed


Artificial Intelligence, Chemometrics, Content Uniformity, Dissolution, Formulation, NIR, Near Infra-red, Near Infrared Spectroscopy (NIRS), PAT, Process Analytical Technology, PC, Principal Component, PCA, Principal Component Analysis, PLS, Partial Least Square, QbD, Quality by Design, Quality by Design (QbD), RMSEP, Root Mean Square Error of Prediction, SEM, Structural Equation Modelling

Radiology Radiology

Convolutional neural network-based automated segmentation and labeling of the lumbar spine X-ray.

In Journal of craniovertebral junction & spine

Purpose : This study investigated the segmentation metrics of different segmentation networks trained on 730 manually annotated lateral lumbar spine X-rays to test the generalization ability and robustness which are the basis of clinical decision support algorithms.

Methods : Instance segmentation networks were compared to semantic segmentation networks based on different metrics. The study cohort comprised diseased spines and postoperative images with metallic implants.

Results : However, the pixel accuracies and intersection over union are similarly high for the best performing instance and semantic segmentation models; the observed vertebral recognition rates of the instance segmentation models statistically significantly outperform the semantic models' recognition rates.

Conclusion : The results of the instance segmentation models on lumbar spine X-ray perform superior to semantic segmentation models in the recognition rates even by images of severe diseased spines by allowing the segmentation of overlapping vertebrae, in contrary to the semantic models where such differentiation cannot be performed due to the fused binary mask of the overlapping instances. These models can be incorporated into further clinical decision support pipelines.

Kónya Sándor, Natarajan Tr Sai, Allouch Hassan, Nahleh Kais Abu, Dogheim Omneya Yakout, Boehm Heinrich

Convolutional neural networks, X-ray, deep neural networks, instance segmentation, lumbar vertebrae, machine learning, postoperative image analysis, semantic segmentation