Reduction of the inappropriate ICD therapies by implementing a new fuzzy logic-based diagnostic algorithm

Area of Science:

• Biomedical Engineering
• Cardiology
• Artificial Intelligence in Medicine

Background:

• Implantable cardioverter-defibrillators (ICDs) are crucial for managing life-threatening arrhythmias.
• Inappropriate therapies, often caused by misclassification of heart rhythms, remain a significant clinical challenge.
• Existing ICD algorithms rely on fixed thresholds, which can limit their accuracy in complex cases.

Purpose of the Study:

• To evaluate a novel fuzzy logic-based detection algorithm (FA) for arrhythmia classification.
• To compare the FA's performance against current ICD arrhythmia detection algorithms.
• To determine if the FA can reduce the rate of inappropriate ICD therapies.

Main Methods:

• A fuzzy logic algorithm utilizing 15 fuzzy rules was developed based on RR intervals, onset, and stability.
• The algorithm was tested on 172 RR recordings from 135 patients, classifying rhythms into 6 categories: VF, VT, ST, DAI, ATF, and NT.
• Performance was compared to existing ICD diagnostic algorithms using data from ICD memory.

Main Results:

• The FA demonstrated high accuracy, correctly diagnosing all cases where inappropriate therapies were administered by existing ICDs (38 cases).
• The incidence of inappropriate therapies was significantly lower with the FA (3 cases) compared to standard ICD diagnosis (38 cases) (p < 0.05).
• The FA achieved superior sensitivity (100%) and specificity (97.8%) compared to tested ICDs (72.9%) for specific diagnostic categories (p < 0.05).

Conclusions:

• The fuzzy logic-based algorithm shows promising diagnostic performance for arrhythmia detection in ICDs.
• Implementation of this FA could significantly decrease the rate of inappropriate ICD therapies.
• The FA proves particularly useful in accurately diagnosing sinus tachycardia, atrial fibrillation, and artifacts, outperforming current ICDs.