Description:Figure 1 illustrates the overall architecture of the proposed intelligent AI-based mental health prediction and counseling system. It shows how the sensors and processing modules are logically connected and how priority-based decision-making is implemented. The system integrates user interaction modules, machine learning-based predictive layers, and a conversational BERT-powered chatbot, forming an end-to-end intelligent assistance framework for student well-being.
The (1) represents the user interface through which students access the system via web or mobile applications. It is connected to the (2) data acquisition layer where standardized inputs from users (survey responses or sensor data) are collected.
The (3) preprocessing module prepares the data for analysis, removing noise and standardizing inputs. This is passed to the (4) feature extraction layer, which identifies significant patterns such as stress indicators or mood fluctuations.
A central processing unit, marked as (5), represents the Random Forest-based prediction engine that analyzes the features and categorizes the user's mental health status (e.g., stressed, anxious, normal). The output is passed to the (6) decision-making and recommendation layer which determines if further action is needed such as suggesting relaxation tips or notifying a counselor.
The BERT-based chatbot interface is labeled (7) and receives the classification results to tailor its interaction with the user. It communicates using contextually accurate and empathetic responses, dynamically generated based on the user's inputs.
All system logs and user data are stored in the (8) data storage unit, ensuring that the history of interactions and predictions is available for longitudinal tracking and improvement of the model.
II. Figure 2. Architecture of BERT Algorithm
Figure 2 depicts the internal working of the BERT (Bidirectional Encoder Representations from Transformers) algorithm used in the chatbot module of the proposed system. The flow begins with tokenized input from the user query and moves through the transformer layers for understanding the contextual meaning.
The (1) indicates input embeddings, where each word from the user’s input is tokenized and embedded with positional and segment information.
The (2) block refers to the stack of transformer encoders, each utilizing self-attention mechanisms to extract semantic relationships from the input.
The resulting contextual representations are passed to the (3) contextual output vectors, which contain the semantic meaning of each word in the given context.
This is fed into the (4) classification layer that identifies the intent of the user input (e.g., emotional query, stress-related concern, or academic stress).
The (5) denotes the response generator that composes a natural language reply using predefined templates and dynamically generated content.
The (6) is the feedback module that continuously updates the training data based on counselor reviews and student feedback, thereby improving future conversations.
The (7) is the system’s internal knowledge base, containing mental health resources, FAQs, and coping strategies which the chatbot can draw from when generating responses.
Finally, the (8) represents the conversation manager, which maintains dialogue context, enabling the chatbot to hold a consistent and coherent multi-turn conversation with the student. , Claims:We Claim:
1. An artificial intelligence system for the provision of mental health evaluation and customized counseling to students, consisting of:
 A user interface available through mobile and web-based platforms, allowing students to enter natural language messages;
 A refined Bidirectional Encoder Representations from Transformers (BERT) model customized to process and analyze the natural language inputs in real-time;
 A classification module that scans linguistic patterns, sentiment, and emotional tone to detect signs of stress, anxiety, and other related mental health issues;
 A response generation module that delivers context-relevant responses and interventions based on the detected psychological state, such as tailored well-being recommendations and references to institutional support services;
 A feedback mechanism that enables the system to learn and refine its responses over time through user interaction.
2. The system of claim 1, where the BERT model is fine-tuned on a dataset of student interactions and mental health indicators to improve the accuracy of psychological state classification.
3. The system of claim 1, where the user interface has features that provide confidentiality and anonymity to promote open communication by students.
4. The system of claim 1, where the response generation module is designed to escalate serious cases by alerting prescribed mental health practitioners or support services within the institution.
5. The system of claim 1, where the feedback mechanism incorporates machine learning processes to continually retrain the BERT model and response generation approach using new information and user input.
6. A system for delivering mental health screening and individualized counseling to students through the system of claim 1, including:
 Receiving natural language input from a student through the user interface.
 Determining and interpreting the input with the fine-tuned BERT model.
 Evaluating the processed input to determine whether there are signs of stress, anxiety, or other related mental health disorders.
• Producing and providing context-dependent responses and interventions based on the evaluation.