A Novel Sensor System for In Vivo Perception Reconstruction Based on Long Short-Term Memory Networks

Area of Science:

• Biomedical Engineering
• Medical Devices
• Artificial Intelligence in Medicine

Background:

• Long-term in vivo pressure monitoring is crucial for medical insights and patient care.
• Current artificial anal sphincters lack sensory feedback, risking tissue necrosis and patient discomfort.
• Perception reconstruction is vital for managing conditions like fecal incontinence.

Purpose of the Study:

• To develop a novel self-packaging strain gauge sensor system for in vivo perception reconstruction.
• To enhance the prediction accuracy of intestinal content amount using artificial intelligence.
• To validate the robustness and efficacy of the sensor system in real-world conditions.

Main Methods:

• Design and implementation of a self-packaging strain gauge sensor system for in vivo use.
• Application of long short-term memory (LSTM) networks for time-series data analysis and prediction.
• Comparative analysis with linear regression (LR) using in vitro and in vivo experimental data.

Main Results:

• The LSTM model demonstrated a significant improvement in prediction accuracy, reducing Root-Mean-Square Error (RMSE) by over 69% compared to LR.
• In vivo experiments confirmed the robustness of the novel sensor system, even with limited data.
• The system successfully enabled accurate prediction of intestinal content amount for improved perception reconstruction.

Conclusions:

• The proposed self-packaging sensor system integrated with LSTM networks offers a highly accurate and robust solution for in vivo pressure monitoring.
• This technology has the potential to significantly improve patient outcomes in managing fecal incontinence and other conditions requiring sensory feedback.
• The study highlights the effectiveness of AI in enhancing the capabilities of implantable medical devices.