Multi-labeled neural network model for automatically processing cardiomyocyte mechanical beating signals in drug...

Area of Science:

• Biomedical Engineering
• Pharmacology
• Artificial Intelligence in Drug Discovery

Background:

• High-throughput cardiotoxicity assessment is crucial for preclinical drug development to prevent drug-induced adverse effects.
• Current methods for cardiotoxicity evaluation lack efficiency and intelligence, necessitating advanced platforms.
• Automatic and intelligent drug assessment platforms are in high demand for preclinical cardiotoxicity investigations.

Purpose of the Study:

• To develop an automatic and intelligent strategy for assessing drug-induced cardiotoxicity using cardiomyocytes mechanical beating signals.
• To classify different drugs based on their cardiotoxic effects and predict the severity of cardiotoxicity.
• To improve the efficiency and accuracy of preclinical drug screening for cardiotoxicity.

Main Methods:

• An automatic feature extraction technique was combined with a multi-labeled neural network (MLNN).
• Cardiomyocytes mechanical beating signals were detected using an interdigital electrode biosensor.
• The MLNN processed bio-signal data for classification and prediction of cardiotoxicity.

Main Results:

• The MLNN achieved high accuracy (up to 96%) and AUC (over 98%) in classifying drug-induced cardiotoxicities.
• A high correlation (over 0.90) was observed between MLNN-predicted and actual drug concentrations, indicating accurate severity prediction.
• The intelligent strategy successfully classified cardiotoxic effects and predicted drug concentrations, outperforming conventional methods.

Conclusions:

• The proposed automatic and intelligent strategy demonstrates significant potential for identifying and classifying drug-induced cardiotoxicity.
• This novel bio-signal processing algorithm offers a promising approach for cardiological and pharmaceutical applications in drug safety.
• The system enhances preclinical drug development by providing efficient and accurate cardiotoxicity assessment.