Feature selection is important in data representation and intelligent diagnosis.Elastic net is one of the most widely used feature selectors. However, the features selected are dependant on the training data, and their weights dedicated for regularized regression are irrelevant to their importance if used for feature ranking, that degrades the model interpretability and extension. In this study, an intuitive idea is put at the end of multiple times of data splitting and elastic net based feature selection. It concerns the frequency of selected features and uses the frequency as an indicator of feature importance. After features are sorted according to their frequency, linear support vector machine performs the classification in an incremental manner. At last, a compact subset of discriminative features is selected by comparing the prediction performance. Experimental results on breast cancer data sets (BCDR-F03, WDBC, GSE 10810, and GSE

In order to improve the accuracy and precision of music generation assisted by robotics, this study analyzes the application of deep learning in piano music generation. Firstly, based on the basic concepts of robotics and deep learning, the advantages of long short-term memory (LSTM) networks are introduced and applied to the piano music generation. Meanwhile, based on LSTM, dropout coefficients are used for optimization. Secondly, various parameters of the algorithm are determined, including the effects of the number of iterations and neurons in the hidden layer on the effect of piano music generation. Finally, the generated music sequence spectrograms are analyzed to illustrate the accuracy and rationality of the algorithm. The spectrograms are compared with the music sequence spectrograms generated by the traditional restricted Boltzmann machine (RBM) music generation algorithm. The results show that (1) when the dropout coefficient value is 0.7, the function converges faster, and the experimental results are better; (2) when the number of iterations is 6000, the error between the generated music sequence and the original music is the smallest; (3) the number of hidden layers of the network is set to 4. When the number of neurons in each hidden layer is set to 1024, the training result of the network is optimal; (4) compared with the traditional RBM piano music generation algorithm, the LSTM-based algorithm and the sampling frequency distribution tend to be consistent with the original sample. The results show that the network has good performance in music generation and can provide a certain reference for automatic music generation.

Freezing of gait (FOG) greatly impacts the daily life of patients with Parkinson’s disease (PD). However, predictors of FOG in early PD are limited. Moreover, recent neuroimaging evidence of cerebral morphological alterations in PD is heterogeneous. We aimed to develop a model that could predict the occurrence of FOG using machine learning, collaborating with clinical, laboratory, and cerebral structural imaging information of early drug-naïve PD and investigate alterations in cerebral morphology in early PD. Data from 73 healthy controls (HCs) and 158 early drug-naïve PD patients at baseline were obtained from the Parkinson’s Progression Markers Initiative cohort. The CIVET pipeline was used to generate structural morphological features with T1-weighted imaging (T1WI). Five machine learning algorithms were calculated to assess the predictive performance of future FOG in early PD during a 5-year follow-up period. We found that models trained with structural morphological features showed fair to good performance (accuracy range, 0.67–0.73). Performance improved when clinical and laboratory data was added (accuracy range, 0.71–0.78). For machine learning algorithms, elastic net-support vector machine models (accuracy range, 0.69–0.78) performed the best. The main features used to predict FOG based on elastic net-support vector machine models were the structural morphological features that were mainly distributed in the left cerebrum. Moreover, the bilateral olfactory cortex (OLF) showed a significantly higher surface area in PD patients than in HCs. Overall, we found that T1WI morphometric markers helped predict future FOG occurrence in patients with early drug-naïve PD at the individual level. The OLF exhibits predominantly cortical expansion in early PD.

No abstract

Writing is a recently acquired skill to human behavioral repertoire, essential in industrialized societies. In the clinic, writing impairment is evident in one-third of stroke patients. This study aimed to find out the cognitive features that contribute to writing impairment of stroke patients in two different writing systems (logographic and phonological). Cognitive profiles were assessed using the Birmingham Cognitive Screen in two cohorts, China (244 patients) and UK (501 patients). The datasets were analyzed separately using an identical procedure. Elastic net was used to rank the importance of different cognitive abilities (features) to writing skill; and linear support vector machine was used to identify the discriminative features needed to accurately identify the stroke patients with and without writing impairments. The prediction performance was evaluated with the area under the curve (AUC), accuracy (ACC), sensitivity (SEN), and specificity (SPE). For the China cohort, writing numbers, complex figure copy, and number calculation obtained good prediction performance on writing impairments with AUC 0.85 ± 0.06, ACC (89 ± 3) %, SEN (81 ± 10) %, and SPE (90 ± 27) %. Concerning the UK data, writing numbers, number calculation, non-word reading, and auditory sustain attention achieved AUC 0.79 ± 0.04, ACC (83 ± 3) %, SEN (74 ± 9) %, and SPE (84 ± 3) %. A small number of patients in both cohorts (China: 9/69, UK: 24/137), who were impaired in writing, were consistently misclassified. Two patients, one in each cohort, showed selective impairments in writing, while all remaining patients were impaired in attention, language, and/or praxis tasks. The results showed that the capability to write numbers and manipulate them were critical features for predicting writing abilities across writing systems. Reading abilities were not a good predictor of writing impairments across both cohorts. Constructive praxis (measured by complex figure copy) was relevant to impairment classification in characters-based writing (China), while phonological abilities (measured by non-word reading) were important features for impairment prediction in alphabetic writing (UK). A small proportion minority of cases with writing deficits were related to different impairment profiles. The findings in this study highlight the multifaceted nature of writing deficits and the potential use of computation methods for revealing hidden cognitive structures in neuropsychological research.