Combating mental illnesses such as depression and anxiety has become a global concern. As a result of the necessity for finding effective ways to battle these problems, machine learning approaches have been included in healthcare systems for the diagnosis and probable prediction of the treatment outcomes of mental health conditions. With the growing interest in machine and deep learning methods, analysis of existing work to guide future research directions is necessary. In this study, 33 articles on the diagnosis of schizophrenia, depression, anxiety, bipolar disorder, post-traumatic stress disorder (PTSD), anorexia nervosa, and attention deficit hyperactivity disorder (ADHD) were retrieved from various search databases using the preferred reporting items for systematic reviews and meta-analysis (PRISMA) review methodology. These publications were chosen based on their use of machine learning and deep learning technologies, individually assessed, and their recommended methodologies were then classified into the various disorders included in this study. In addition, the difficulties encountered by the researchers are discussed, and a list of some public datasets is provided.

Detecting depression involves using standardized questionnaires like the Patient Health Questionnaires (PHQ-8/9). Yet, patients might not always provide genuine responses, leading to potential misdiagnoses. Therefore, the need for a means to detect depression in patients without the use of preset questions is of high importance. Addressing this challenge, our study aims to discern telltale symptoms from statements made by the patient. We harness both audio and text data, proposing an Additive cross-modal attention network to learn and pick up the appropriate weights that best capture the cross-modal interactions and relationships between both features using BiLSTM as the backbone of both modalities. We tested our approach on the DAIC-WOZ dataset for depression detection and also evaluated our model performance on the EATD-Corpus. Benchmarked against similar studies on this dataset, our method demonstrates commendable efficacy in both classification and regression models for both unimodal and multimodal approaches. Our findings underscore the potential of our model to effectively detect depression in patients while using textual and speech modalities without the necessary use of preset questions for effective detection.

Automatic pain recognition is essential in healthcare. In previous studies, automatic pain recognition methods preferentially apply the features extracted from physiological signals for conventional models. This method provides good performance but mainly relies on medical expertise for feature extraction of physiological signals. This paper presents a deep learning approach based on physiological signals that have the role of both feature extraction and classification, regardless of medical expertise. We propose multi-level context information for each physiological signal discriminating between pain and painlessness. Our experimental results prove that multi-level context information has more significant performances than uni-level context information based on Part A of the BioVid Heat Pain database and the Emopain 2021 dataset. For Part A of the BioVid Heat Pain database, our experimental results for pain recognition tasks include: Pain 0 and Pain 1, Pain 0 and Pain 2, Pain 0 and Pain 3, and Pain 0 and Pain 4. In the classification task between Pain 0 and Pain 4, the results achieve an average accuracy of 84.8 ± 13.3% for 87 subjects and 87.8 ± 11.4% for 67 subjects in a Leave-One-Subject-Out cross-validation evaluation. The proposed method adopts the ability of deep learning to outperform conventional methods on physiological signals.

The human brain remains continuously active, whether an individual is working or at rest. Mental activity is a daily process, and if the brain becomes excessively active, known as overload, it can adversely affect human health. Recently, advancements in early prediction of mental health conditions have emerged, aiming to prevent serious consequences and enhance the overall quality of life. Consequently, the estimation of mental status has garnered significant attention from diverse researchers due to its potential benefits. While various signals are employed to assess mental state, the electroencephalogram, containing extensive information about the brain, is widely utilized by researchers. In this paper, we categorize mental workload into three states (low, middle, and high) and estimate a continuum of mental workload levels. Our method leverages information from multiple spatial dimensions to achieve optimal results in mental estimation. For the time domain approach, we employ Temporal Convolutional Networks. In the frequency domain, we introduce a novel architecture based on combining residual blocks, termed the Multi-Dimensional Residual Block. The integration of these two domains yields significant results compared to individual estimates in each domain. Our approach achieved a 74.98% accuracy in the three-class classification, surpassing the provided data results at 69.00%. Specially, our method demonstrates efficacy in estimating continuous levels, evidenced by a corresponding Concordance Correlation Coefficient (CCC) result of 0.629. The combination of time and frequency domain analysis in our approach highlights the exciting potential to improve healthcare applications in the future.