The correct sitting posture in a wheelchair is crucial for paralyzed people. This helps prevent problems such as pressure ulcers, muscle contractures, and respiratory problems. A paralyzed person with poor sitting posture is highly likely to slip out of their wheelchair. To prevent this from happening and consistently maintain paralyzed individuals under observation, a new model, the Emperor Penguin Optimized Sensor-Infused Wheelchair (EPIC), has been designed to monitor the position and health of the individual in the wheelchair in real-time. A Force Sensitive Resistor (FSR) sensor and an ultrasonic sensor continuously transmit information to the Arduino UNO R4 Wi-Fi board. The Emperor Penguin Optimizer Algorithm (EPOA) was used to select the features sent from the Arduino board to the ESP8266-Wi-Fi module. A Deep Maxout Network (DMN) was used to predict the posture of a wheelchair-using patient following the feature selection phase. A mobile application for Android collects data from the ESP32 module and estimates posture to inform the caretaker about the user's current posture and health status. Evaluation metrics such as precision, accuracy, sensitivity, and specificity have been used to determine the efficiency in the EPIC framework, which improves overall accuracy by 10.1%, 7.73%, and 2.84% for better posture recognition.
Keywords: Deep Maxout Network; Deep learning; Emperor Penguin Optimizer; Feature selection; Healthcare monitoring system; Sensor networks.
© 2025. The Author(s).