Techtransfer Case 1: New analysis algorithms for the evaluation of electrical signals generated by living cells
Within the PROMOS project, one of the key technology transfer objectives is the development of advanced analytical tools that can translate complex biological data into actionable insights. TT Case 1 addresses a central scientific and technological challenge: the precise interpretation of electrical activity generated by human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) cultured on multi-electrode arrays (MEAs).
These cells reproduce many of the electrophysiological properties of the human heart. When exposed to chemical compounds, they respond with measurable changes in their field potentials. Such responses hold crucial information on drug-induced cardiotoxicity but the sheer volume and complexity of MEA data make manual evaluation time-consuming, subjective, and difficult to scale. Developing reliable and automated analysis methods is therefore essential to advancing preclinical safety screening.
To meet this challenge, PROMOS partners FH Kärnten (AT) and Eurac Research (IT) have designed a new software pipeline incorporating three deep neural networks (DNNs), each tailored to a specific analytical step:
- Depolarization spike localization,
- T-wave detection, and
- Arrhythmia classification.
Together, these models reconstruct the temporal structure of the cardiac cycle and identify irregularities with high precision. Their training requires an extensive training dataset. Several thousand MEA signal sequences are being manually annotated to generate robust ground-truth labels, which is a significant undertaking that ensures high-fidelity downstream predictions.
This activity brings together complementary strengths from the consortium: expertise in MEA-based electrophysiology, biomedical data processing, and state-of-the-art AI model development. The outcome is a scalable, replicable analytical tool capable of supporting:
- Cardiotoxicity assessment of established and emerging compounds,
- Systematic comparison of pharmacological effects across experimental conditions,
- Improved reproducibility in high-content cardiac safety screening.
By integrating these algorithms into a user-friendly software platform, the project provides researchers with a powerful resource for understanding drug effects on human cardiac cells. The approach exemplifies how cross-border collaboration and multidisciplinary innovation can accelerate the transfer of scientific advances into practical applications with clear biomedical relevance.