Scholarlink

<p>Significant health hazards have been associated with the COVID-19 pandemic, especially for women, whose mental and physical health have suffered greatly. In order to predict the risk of COVID-19 in women, this study investigates the use of stacked artificial neural networks (ANNs) in hidden layers with a proposed Twish activation function. The study incorporates many psychological and health-related measures into the artificial neural network framework, facilitating an all-encompassing evaluation of risk. Important measures include the Beck Anxiety Inventory (BAI), which assesses anxiety specifically related to pandemics (ANN1), the Social Readjustment Rating Scale (SRRS), which assesses stress from significant life events (ANN2), the Centre for Epidemiologic Studies Depression Scale (CES-D), which assesses symptoms of depression (ANN3), the Charlson Comorbidity Index (CCI), which assesses the impact of pre-existing health conditions (ANN4), and the Prenatal Distress Questionnaire (PDQ), which measures stress in pregnant women during the pandemic (ANN5).The ANN model enhances performance in the hidden layers by using the Twish activation function, enabling more precise and effective COVID-19 risk prediction. Predicting risk variables for women is especially helpful because of the intricacy of mental health, co-occurring conditions, and stress associated with pregnancy during the pandemic. All classes reveal that the proposed model is remarkably accurate, and several of its modifications produce outstanding outcomes. With a 97.75% accuracy rate, ANN1 has the strongest predicting capability. With accuracy scores of 97.50%, ANN2 and ANN4 continuously demonstrate good performance, highlighting dependable performance. ANN3, which has an accuracy of 98.00%, draws attention to outstanding performance. Furthermore, with an accuracy rate of 97.50%, ANN5 retains great predictive capabilities, demonstrating good and consistent performance in a variety of circumstances.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

TWISH: an enhanced activation function for optimizing mental health and comorbidity risk assessment during COVID-19 using deep learning

  • 期刊论文

  • 发表日期：

    2025年2月18日

查看全文
  • Usharani Bhimavarapu

摘要

Significant health hazards have been associated with the COVID-19 pandemic, especially for women, whose mental and physical health have suffered greatly. In order to predict the risk of COVID-19 in women, this study investigates the use of stacked artificial neural networks (ANNs) in hidden layers with a proposed Twish activation function. The study incorporates many psychological and health-related measures into the artificial neural network framework, facilitating an all-encompassing evaluation of risk. Important measures include the Beck Anxiety Inventory (BAI), which assesses anxiety specifically related to pandemics (ANN1), the Social Readjustment Rating Scale (SRRS), which assesses stress from significant life events (ANN2), the Centre for Epidemiologic Studies Depression Scale (CES-D), which assesses symptoms of depression (ANN3), the Charlson Comorbidity Index (CCI), which assesses the impact of pre-existing health conditions (ANN4), and the Prenatal Distress Questionnaire (PDQ), which measures stress in pregnant women during the pandemic (ANN5).The ANN model enhances performance in the hidden layers by using the Twish activation function, enabling more precise and effective COVID-19 risk prediction. Predicting risk variables for women is especially helpful because of the intricacy of mental health, co-occurring conditions, and stress associated with pregnancy during the pandemic. All classes reveal that the proposed model is remarkably accurate, and several of its modifications produce outstanding outcomes. With a 97.75% accuracy rate, ANN1 has the strongest predicting capability. With accuracy scores of 97.50%, ANN2 and ANN4 continuously demonstrate good performance, highlighting dependable performance. ANN3, which has an accuracy of 98.00%, draws attention to outstanding performance. Furthermore, with an accuracy rate of 97.50%, ANN5 retains great predictive capabilities, demonstrating good and consistent performance in a variety of circumstances.