Automatic neonatal sleep stage classification: A comparative study

Reducing Noise in the NICU

BACKGROUND

In the neonatal intensive care unit (NICU), elevated noise negatively impacts the neurodevelopmental environment, interrupts sleep, and can affect brain development in neonates. The American Academy of Pediatrics recommends that noise levels in the NICU should not exceed 45 dB.

PURPOSE

The project aims were to: (1) decrease average noise level by 10% from baseline and (2) decrease exposure to severe noise (>65 dB) to <5% of the time.

METHODS

This quality improvement project was conducted during 2021-2022 as a pre/post observational design in a Level IV NICU in New York City. We monitored sound levels for 20-24 h, 5 d/wk. Quality improvement interventions included: novel approaches to staff education, visual cues for when noise thresholds were exceeded, parent education, including access to personal decibel meters, technical improvements to vital sign monitors and entry doors, and defined quiet times (HUSH) for 2 h each 12-hour shift.

RESULTS

Education efforts and technical improvements successfully reduced median noise levels within the stepdown unit (P < .001), though not in the acute care NICU. In contrast, the implementation of 2-hour periods of enforced "quiet time" every 12 h effectively reduced both median noise levels and the incidence of severe noise (>65 dB) in both locations.

IMPLICATIONS FOR PRACTICE AND RESEARCH

The HUSH strategy may be a sustainable way to decrease noise in the NICU. Future projects should prioritize education and dedicated quiet times to align with recommended standards, while research should explore the long-term developmental impacts of excessive noise levels on neonatal growth.

Connecting the dots: Exploring brain connectivity during responsibility recognition in construction contract negotiations

Despite recent advancements in monitoring brain activity, causal relationships within the brain during responsibility identification in construction contracts remain unexplored. We aimed to understand the neural mechanisms involved in the cognitive components and their interactions related to contract text reading by delving into the brain mechanisms of contract responsibility identification. This study investigated students' brain connectivity using electroencephalography (EEG) data during a text-based contract responsibility-identification task. It employed an adaptive directed transfer function based on Granger causality to simulate directed and time-varying information flow in observed brain activity. We evaluated the EEG records of 18 participants under two reading conditions (involving or not involving contractor responsibility). During responsibility identification, the most substantial information exchange occurs in the somatosensory area of the brain. The results revealed a "top-down" cortical mechanism for responsibility identification, with the left parietal-occipital area (PO3) as the central hub promoting connectivity structures. These findings indicate that the perceptual processing of contract responsibility texts is associated with higher visual learning and memory quality. Contracts without contractor-responsibility clauses resulted in more substantial information flow output in the frontal cortex and consumed more cognitive resources. Our findings advance the understanding of cognitive processes involved in contract responsibility identification, providing a framework for investigating causal relationships within the brain and novel insights into cortical mechanisms. By identifying the neural basis of responsibility identification, stakeholders can develop effective training programs for negotiators and enhance their ability to interpret and implement construction contracts.

Improved neural network with multi-task learning for Alzheimer's disease classification

Alzheimer's disease(AD) poses a significant challenge due to its widespread prevalence and the lack of effective treatments, highlighting the urgent need for early detection. This research introduces an enhanced neural network, named ADnet, which is based on the VGG16 model, to detect Alzheimer's disease using two-dimensional MRI slices. ADNet incorporates several key improvements: it replaces traditional convolution with depthwise separable convolution to reduce model parameters, replaces the ReLU activation function with ELU to address potential issues with exploding gradients, and integrates the SE(Squeeze-and-Excitation) module to enhance feature extraction efficiency. In addition to the primary task of MRI feature extraction, ADnet is simultaneously trained on two auxiliary tasks: clinical dementia score regression and mental state score regression. Experimental results demonstrate that compared to the baseline VGG16, ADNet achieves a 4.18% accuracy improvement for AD vs. CN classification and a 6% improvement for MCI vs. CN classification. These findings highlight the effectiveness of ADnet in classifying Alzheimer's disease, providing crucial support for early diagnosis and intervention by medical professionals. The proposed enhancements represent advancements in neural network architecture and training strategies for improved AD classification.

Classification of EEG Signals Based on Sparrow Search Algorithm-Deep Belief Network for Brain-Computer Interface

In brain-computer interface (BCI) systems, challenges are presented by the recognition of motor imagery (MI) brain signals. Established recognition approaches have achieved favorable performance from patterns like SSVEP, AEP, and P300, whereas the classification methods for MI need to be improved. Hence, seeking a classification method that exhibits high accuracy and robustness for application in MI-BCI systems is essential. In this study, the Sparrow search algorithm (SSA)-optimized Deep Belief Network (DBN), called SSA-DBN, is designed to recognize the EEG features extracted by the Empirical Mode Decomposition (EMD). The performance of the DBN is enhanced by the optimized hyper-parameters obtained through the SSA. Our method's efficacy was tested on three datasets: two public and one private. Results indicate a relatively high accuracy rate, outperforming three baseline methods. Specifically, on the private dataset, our approach achieved an accuracy of 87.83%, marking a significant 10.38% improvement over the standard DBN algorithm. For the BCI IV 2a dataset, we recorded an accuracy of 86.14%, surpassing the DBN algorithm by 9.33%. In the SMR-BCI dataset, our method attained a classification accuracy of 87.21%, which is 5.57% higher than that of the conventional DBN algorithm. This study demonstrates enhanced classification capabilities in MI-BCI, potentially contributing to advancements in the field of BCI.