Hybrid Intelligent System to Perform Fault Detection on BIS Sensor During Surgeries.

Area of Science:

• Anesthesiology
• Biomedical Engineering
• Artificial Intelligence in Medicine

Background:

• Patient monitoring devices are crucial for drug infusion during surgery.
• Faults in hypnotic sensors, like the Bispectral Index (BIS) monitor, can compromise patient safety.
• Accurate assessment of unconsciousness level is vital for effective general anesthesia.

Purpose of the Study:

• To develop and validate a fault detection system for BIS sensors used in propofol-induced general anesthesia.
• To identify and disregard BIS signal disturbances that lack clinical significance during surgery.
• To enhance the reliability of automated or clinician-guided drug dosage calculations.

Main Methods:

• A hybrid intelligent system combining clustering and regression techniques was developed to predict BIS signal behavior.
• The method compares the measured BIS signal against predictions derived from propofol dosage and electromyogram (EMG) signals.
• Validation was performed using a dataset from real surgical cases under general anesthesia.

Main Results:

• The developed system effectively identifies BIS episodes affected by disturbances with null clinical value.
• The fault detection method demonstrated accuracy in real-world surgical scenarios.
• The system successfully distinguishes between reliable and unreliable BIS data, improving monitoring integrity.

Conclusions:

• The proposed fault detection system enhances the safety and reliability of hypnotic sensor monitoring during general anesthesia.
• By filtering out erroneous BIS data, the system supports more accurate drug dose calculations.
• This research contributes to safer surgical procedures through improved patient monitoring technology.