Mental health is just as important as physical health. Globally, there is a growing concern due to the rise of mental health problems. Mental health problems impair individuals’ ability to think clearly, behave responsively, or express themselves. The recent pandemic has led to a spike in people using social media to express themselves. Data from these social media platforms can be used to learn more about users’ mental states and determine the causes of mental health problems. However, due to the variability and complexity of users’ language, it is very challenging for conventional machine learning and deep learning models to identify these causes. In this research, we propose a novel sentence-level analysis framework based on a hybrid deep learning model to overcome these challenges. We evaluated our model’s efficacy using data from the social media platform Reddit, and our model outperforms several baseline models. Our research findings provide a new perspective on identifying the causes behind mental health issues and would help mental health professionals develop better diagnoses and treatments for patients.

Post-stroke patient rehabilitation involves physical therapy primarily to re-learn activities due to the damage caused by stroke. The patient performs these therapies in the in-patient clinical care center under the direct supervision of clinicians to monitor the progress. However, these therapy services are costly and limited availability due to the lack of clinicians, constraint dependent (e.g., clinic services only). With technological advancement and fast-growing artificial intelligence (AI), human movement can be captured through skeletal joint movement using various sensors (e.g., Kinect) and develop AI-enabled applications to assess the performed exercise automatically. In this work, we proposed a novel hybrid quantum neural network model to evaluate post-stroke patients’ exercises sensed through Kinect sensors. Our proposed hybrid quantum neural network model comprises traditional neural network layers and quantum layers to extract inherent hidden features uncaptured by neural networks and perform assessment tasks. We showcase the effectiveness and efficacy of our model by evaluating the performance using a publicly available dataset (UI-PRMD) that consists of ten exercises from 10 users. Our proposed model achieves RMSE error on average 0.044341 and outperforms the traditional machine learning algorithms.

Suicide has become a significant cause of concern worldwide over recent years. The early identification and providing treatment of individuals having suicidal tendencies are necessary for preventing suicides. Past suicidal behavior information of an individual is recorded in the electronic health records (EHR) reports which can help to understand a patient’s current mental health condition. In this paper, to identify the people who are ideating and are anticipating attempting suicide, we propose a novel model named CMTN, which utilizes the textual EHR data for the prediction of suicidal behaviors. The proposed framework employs convolutional and transformer layers to capture local and global relationships in the text and the attention mechanism to assess the significance of various input text components. Overall, the suggested model has achieved the highest precision for the SA class with a score of 0.97 and the highest recall and f1-score of 0.56 and 0.52, respectively, for the SI class, compared with all other state-of-the-art and baseline models. We have also employed different embeddings such as BERT, BioBERT, and PubMedBERT to our state-of-the-art model and illustrated the model’s improved performance. In addition, we have also shared the data alignment and annotation extraction algorithms in this paper, allowing other researchers to generate the dataset, thereby expediting development in the prevention of suicides.