Early Life Stress Detection Using Physiological Signals and Machine Learning Pipelines

Abstract

Pregnancy and early childhood are two vulnerable times when immunological plasticity is at its peak and exposure to stress may substantially raise health risks. However, to separate the effects of adversity during vulnerable times of the lifetime from those across the entire lifespan, we require deeper phenotyping. Stress is one of the challenges which everyone can face with this issue. It is a type of feeling which contains mental pressure and comes from daily life matters. There are many research and investments regarding this problem to overcome or control this complication. Pregnancy is a susceptible period for the child and the mother taking stress can affect the child’s health after birth. The following matter can happen based on natural disasters, war, death or separation of parents, etc. Early Life Stress (ELS) has a connection with psychological development and metabolic and cardiovascular diseases. In the following research, the main focus is on Early Life Stress control during pregnancy of a healthy group of women that are at risk of future disease during their pregnancy. This study looked at the relationship between retrospective recollections of childhood or pregnancy hardship and inflammatory imbalance in a group of 53 low-income, ethnically diverse women who were seeking family-based trauma treatment after experiencing interpersonal violence. Machine learning Convolutional Neural Networks (CNNs) are applied for stress detection using short-term physiological signals in terms of non-linear and for a short term. The focus concepts are heart rate, and hand and foot galvanic skin response.

Early Mental Stress Detection Using Q-Learning Embedded Starling Murmuration Optimiser-Based Deep Learning Model

“…This equates to the fact that each iteration of the population will have to go through the KNN classification algorithm. To circumvent such issues in the process of feature selection, we propose a variance based feature selection technique as shown in equation (11) that uses the data variance as a tool to determine the fitness value. The…”

“…By categorising states and actions in this way, QLESMO can efficiently explore the search space and handle a wide range of real-world problems with robustness. The pseudocode of QLESMO algorithm combined with the proposed feature selection methodology as the fitness function evaluator, equation (11), is given in Algorithm 3.…”

“…The data generated from wearable sensors and social media can be analysed using advanced analytics and machine learning techniques to identify patterns and predict stress levels. Machine learning algorithms such as Random Forest and Support Vector Machines are being used to develop predictive models for stress detection [11]. Natural language processing (NLP) techniques are being used to analyse text data from social media and emails to identify linguistic indicators of stress.…”

An evolutionary and insightful graph neural network based hybrid model for deciphering tenacious stress detection of humans using facial emotion recognition: coinage to affective computing

“…Self-Report method analysis method assess stress by analyzing factors such as questionaries and surveys, diaries or journal entries (Zainudin et al 2021). However, there are few shortcomings associated with these methods like physiological signal-based methods requires a) Specialized equipment and sensors to measure HRV, EEG and EDA signals, and this complexity and dependency on equipment can limit their practicality and accessibility (Maaoui and Pruski 2018) ; b) These responses can vary significantly between individuals i.e., what may be considered a stress response for one person may not be the same for another, hence this inter-individual variability is challenging in decision making; c) Physiological signals can be influenced by a range of emotions like excitement, fear, or even physical exertion and elicit similar physiological response, making it difficult to differentiate between stress specific patterns from other emotional states (Kushagra Nigam et al 2021); d) These signals may vary depending upon context and environmental factors such as temperature, noise levels, or presence of others can influence physiological response (Katmah et al 2021); e) Methods such as cortisol level analysis, may involve invasive or costly procedures (Shahbazi and Byun 2023). Similarly, in Self-Report Method, few major shortcomings are: a) These methods rely on individual's own perceptions and interpretations of their stress levels (Maltman et al 2023); b) Mainly individual's subjective experiences are captured and this may not provide a complete understanding of the physiological responses associated with stress (Witte et al 2021); c) Stress perception may vary depending on situational context, mood, or personal factors, leading to inconsistent reporting; d) the accuracy and reliability can be compromised when individuals are asked to recall stress experiences from the past, particularly for longterm or chronic stress assessments; e) these methods often rely on periodic assessments or surveys, which may not capture stress fluctuations in real-time (Theon et al 2023), f) the diversity of available questionnaires and scale make it challenging to establish a universally applicable measure for stress detection.…”