Schizo-Net: A novel Schizophrenia Diagnosis Framework Using Late Fusion Multimodal Deep Learning on Electroencephalogram-Based Brain Connectivity Indices

AI-driven early diagnosis of specific mental disorders: a comprehensive study

One of the areas where artificial intelligence (AI) technologies are used is the detection and diagnosis of mental disorders. AI approaches, including machine learning and deep learning models, can identify early signs of bipolar disorder, schizophrenia, autism spectrum disorder, depression, suicidality, and dementia by analyzing speech patterns, behaviors, and physiological data. These approaches increase diagnostic accuracy and enable timely intervention, which is crucial for effective treatment. This paper presents a comprehensive literature review of AI approaches applied to mental disorder detection using various data sources, such as survey, Electroencephalography (EEG) signal, text and image. Applications include predicting anxiety and depression levels in online games, detecting schizophrenia from EEG signals, detecting autism spectrum disorder, analyzing text-based indicators of suicidality and depression, and diagnosing dementia from magnetic resonance imaging images. eXtreme Gradient Boosting (XGBoost), light gradient-boosting machine (LightGBM), random forest (RF), support vector machine (SVM), K-nearest neighbor were designed as machine learning models, and convolutional neural networks (CNN), long short-term memory (LSTM) and gated recurrent unit (GRU) models suitable for the dataset were designed as deep learning models. Data preprocessing techniques such as wavelet transforms, normalization, clustering were used to optimize model performances, and hyperparameter optimization and feature extraction were performed. While the LightGBM technique had the highest performance with 96% accuracy for anxiety and depression prediction, the optimized SVM stood out with 97% accuracy. Autism spectrum disorder classification reached 98% accuracy with XGBoost, RF and LightGBM. The LSTM model achieved a high accuracy of 83% in schizophrenia diagnosis. The GRU model showed the best performance with 93% accuracy in text-based suicide and depression detection. In the detection of dementia, LSTM and GRU models have demonstrated their effectiveness in data analysis by reaching 99% accuracy. The findings of the study highlight the effectiveness of LSTM and GRU for sequential data analysis and their applicability in medical imaging or natural language processing. XGBoost and LightGBM are noted to be highly accurate ML tools for clinical diagnoses. In addition, hyperparameter optimization and advanced data pre-processing approaches are confirmed to significantly improve model performance. The results obtained with this study have revealed the potential to improve clinical decision support systems for mental disorders with AI, facilitating early diagnosis and personalized treatment strategies.

A review of studies on constructing classification models to identify mental illness using brain effective connectivity

Brain effective connectivity (EC) is a functional measurement that reflects the causal effects and topological relationships of neural activities. Recent research has increasingly focused on the classification for mental illnesses and healthy controls using brain EC; however, no comprehensive reviews have synthesized these studies. Therefore, the aim of this review is to thoroughly examine the existing literature on constructing diagnosis model for mental illnesses using brain EC. We first conducted a systematical literature search and thirty-five papers met the inclusion criteria. Subsequently, we summarized the approaches for estimating EC, the classification and validation methods used, the accuracies of models, and the main findings. Finally, we discussed the limitations of current research and the challenges in future research. These summaries and discussion provide references for future research on mental illnesses identification based on brain EC.

EISATC-Fusion: Inception Self-Attention Temporal Convolutional Network Fusion for Motor Imagery EEG Decoding

The motor imagery brain-computer interface (MI-BCI) based on electroencephalography (EEG) is a widely used human-machine interface paradigm. However, due to the non-stationarity and individual differences among subjects in EEG signals, the decoding accuracy is limited, affecting the application of the MI-BCI. In this paper, we propose the EISATC-Fusion model for MI EEG decoding, consisting of inception block, multi-head self-attention (MSA), temporal convolutional network (TCN), and layer fusion. Specifically, we design a DS Inception block to extract multi-scale frequency band information. And design a new cnnCosMSA module based on CNN and cos attention to solve the attention collapse and improve the interpretability of the model. The TCN module is improved by the depthwise separable convolution to reduces the parameters of the model. The layer fusion consists of feature fusion and decision fusion, fully utilizing the features output by the model and enhances the robustness of the model. We improve the two-stage training strategy for model training. Early stopping is used to prevent model overfitting, and the accuracy and loss of the validation set are used as indicators for early stopping. The proposed model achieves within-subject classification accuracies of 84.57% and 87.58% on BCI Competition IV Datasets 2a and 2b, respectively. And the model achieves cross-subject classification accuracies of 67.42% and 71.23% (by transfer learning) when training the model with two sessions and one session of Dataset 2a, respectively. The interpretability of the model is demonstrated through weight visualization method.

Multimodal Assessment of Schizophrenia Symptom Severity From Linguistic, Acoustic and Visual Cues

Assessing the condition of every schizophrenia patient correctly normally requires lengthy and frequent interviews with professionally trained doctors. To alleviate the time and manual burden on those mental health professionals, this paper proposes a multimodal assessment model that predicts the severity level of each symptom defined in Scale for the Assessment of Thought, Language, and Communication (TLC) and Positive and Negative Syndrome Scale (PANSS) based on the patient's linguistic, acoustic, and visual behavior. The proposed deep-learning model consists of a multimodal fusion framework and four unimodal transformer-based backbone networks. The second-stage pre-training is introduced to make each off-the-shelf pre-trained model learn the pattern of schizophrenia data more effectively. It learns to extract the desired features from the view of its modality. Next, the pre-trained parameters are frozen, and the light-weight trainable unimodal modules are inserted and fine-tuned to keep the number of parameters low while maintaining the superb performance simultaneously. Finally, the four adapted unimodal modules are fused into a final multimodal assessment model through the proposed multimodal fusion framework. For the purpose of validation, we train and evaluate the proposed model on schizophrenia patients recruited from National Taiwan University Hospital, whose performance achieves 0.534/0.685 in MAE/MSE, outperforming the related works in the literature. Through the experimental results and ablation studies, as well as the comparison with other related multimodal assessment works, our approach not only demonstrates the superiority of our performance but also the effectiveness of our approach to extract and integrate information from multiple modalities.