TL;DR
This paper introduces a novel hybrid Brain-Computer Interface (BCI) and Internet of Things (IoT) system designed for smart ward environments. By combining electroencephalography (EEG), electrooculography (EOG), and gyroscope data for BCI control, and integrating it with an NB-IoT based monitoring system, the research aims to enhance patient autonomy and ward efficiency, particularly for individuals with mobility impairments.
What Problem Does This Paper Address?
Traditional care environments, especially for an aging population, struggle to provide personalized and efficient care. Manual control systems in wards are often inconvenient and unsuitable for patients with severe physical disabilities (e.g., spinal cord injuries, ALS). While BCIs offer a promising alternative, existing single-mode BCIs often face limitations such as low information transfer rates (ITR), high false operation rates (FOR), or long response times. Furthermore, the integration of patient monitoring and ward management systems needs to be more seamless and intelligent to ensure patient safety and optimize resource allocation. This paper addresses the need for a more robust, accurate, and user-friendly control system for patients, coupled with an efficient IoT-based monitoring infrastructure for smart wards.
Methods
The proposed system is a hybrid BCI and IoT architecture.
Hybrid BCI Control System: * Multimodal Signal Acquisition: The system utilizes EEG (for attention state), EOG (for blinking/clicking), and gyroscope (for cursor movement) signals. * GUI Paradigm: A two-layer GUI is employed. The first layer allows for mode selection and attention switching, while the second layer presents ward-specific control commands. * Cursor Control: Gyroscope data is used for two-dimensional cursor movement on the GUI. * Command Selection: Blinking-related EOG signals are used to "click" or select commands. * Attention-Based Judgment: EEG signals are processed using a Support Vector Machine (SVM) to classify the user's attention state. This attention state, along with cursor position and blink detection, forms a three-condition judgment to execute a command, significantly reducing false operations. * Asynchronous Operation: The system operates asynchronously, meaning users don't need to wait for specific visual cues for command initiation, leading to faster response times.
IoT Monitoring and Management System: * NB-IoT Technology: Narrowband Internet of Things (NB-IoT) is used for efficient and low-power communication of data from various sensors within the ward. * Layered Architecture: The system comprises a perception layer (sensors), a transport layer (NB-IoT chip), and a platform application layer (cloud server for data storage, analysis, and decision-making). * Data Integration: It integrates physiological data (heart rate), environmental data (temperature, humidity, air quality), and positioning data for comprehensive ward management and early warning systems.
Datasets
The study utilized data from nine healthy volunteers (S1-S9) for the BCI experiments. No specific pre-existing datasets were mentioned for training the BCI models; rather, data was collected from participants during the experiments. For the IoT monitoring aspect, environmental sensors and a wearable device were used to collect real-time data.
Key Insights and Ideas
- Hybrid Signal Fusion for Robustness: Combining EEG, EOG, and gyro signals significantly improves BCI performance by leveraging the strengths of each modality. EOG provides intuitive control, gyro offers precise cursor movement, and EEG adds a crucial layer of attention-based validation.
- Attention State as a Gatekeeper: Using EEG to verify the user's attention state before executing an EOG-triggered command is a key innovation for reducing false positives, a common problem in EOG-based BCIs.
- Asynchronous Control for Speed: The asynchronous nature of the BCI, particularly the gyro-based cursor movement, bypasses the need for strict timing associated with visual cues, leading to faster command generation.
- NB-IoT for Efficient Ward Connectivity: The adoption of NB-IoT for the IoT component offers a low-power, cost-effective solution for transmitting diverse ward data, crucial for scalable smart ward implementations.
- Multimodal Monitoring for Comprehensive Safety: Integrating environmental and physiological monitoring provides a holistic view of the patient's well-being and ward conditions, enhancing safety and enabling proactive interventions.
Key Results
- BCI Accuracy: The hybrid BCI system achieved a high accuracy of 96.65 ± 1.44% (with attention detection) and 95.54 ± 1.28% (without attention detection).
- False Operation Rate (FOR): The FOR was significantly reduced to 0.89 ± 0.42 events/min with attention detection, compared to 1.1 ± 0.32 events/min without it.
- Command Response Time (RT): The average response time was 2.65 ± 0.48 s (attention open) and 2.87 ± 0.49 s (attention closed), demonstrating efficient command execution.
- Information Transfer Rate (ITR): The ITR was 53.42 ± 8.44 bits/min (attention open) and 47.43 ± 7.62 bits/min (attention closed), indicating good information throughput.
- IoT Data Transmission Reliability: Network communication tests showed low packet loss rates (0.1-0.2%) and bit error rates (0-0.8%), even under interference, proving the reliability of the NB-IoT system.
- Environmental Monitoring Accuracy: The system demonstrated high accuracy in monitoring environmental parameters, with temperature accuracy within ±1% and humidity within ±2%.
Limitations and Future Work
- Subjectivity of Workload: While workload assessments were conducted, they relied on subjective questionnaires (NASA-TLX), which can have inherent biases.
- Healthy Volunteers: The BCI experiments were conducted with healthy volunteers. Future work should involve testing with patients who have actual physical limitations to fully validate the system's utility in its target environment.
- Fixation Duration and Task Duration: The authors acknowledge that fixation duration and saccade frequency can affect mental load and task efficiency. Future work will focus on optimizing attention detection by finding a reasonable fixation duration and task duration.
- Further Optimization: The authors plan to further improve the hybrid BCI system by optimizing eye-tracking methods and integrating more IoT resources to make the system more user-friendly for disabled individuals.
Summary updated