Image of Person undergoing EEG Scan


Enhancing EEG Data Analysis: Predicting Disconnected Electrode Values using Bidirectional LSTM

Szymon Fonau, CTO @ OctoShrew


This case study presents the development of a novel program aimed at predicting the values of temporarily disconnected EEG electrodes. When patients move or experience disturbances, non-professional grade EEG devices, such as the EMOTIV EEG, often encounter electrode disconnections, resulting in incomplete data. To overcome this challenge, a unique approach leveraging bidirectional LSTM (Long Short-Term Memory) networks was developed. By utilizing data from all other EEG nodes, including recent and new measurements, the program accurately predicts the value of a disconnected electrode. The results of this study provide valuable insights and improve experiment outcomes for our client.


Accurate EEG data plays a vital role in medical and research settings, providing insights into brain activity. However, temporarily disconnected electrodes pose challenges to data integrity and analysis. This case study presents a program that addresses this issue by predicting the values of disconnected EEG electrodes, leveraging bidirectional LSTM networks. By doing so, our client gains better insights and more reliable conclusions from their experiments.


EEG, or Electroencephalography, is a technique used to measure electrical brain activity. Disconnected electrodes can occur due to patient movement or device limitations, leading to gaps in data collection. This case study focuses on non-professional grade EEG devices, which often encounter electrode disconnections, hampering data analysis.

Development Process:

The development process involved collecting and preprocessing EEG data from non-professional grade devices. Instances of temporarily disconnected electrodes were identified. A bidirectional LSTM architecture was implemented, capable of predicting the value of a disconnected electrode by utilizing data from all other EEG nodes, including recent and new measurements. The model was trained and fine-tuned to optimize prediction accuracy, followed by validation and evaluation using unseen data.

Functionality and Features:

The program's key functionality lies in accurately predicting the values of disconnected EEG electrodes. By leveraging bidirectional LSTM networks, the program incorporates data from all other EEG nodes, both recent and new measurements, to provide reliable predictions. This enables continuous monitoring and analysis of brain activity, offering real-time insights for medical professionals and researchers.

Benefits and Impact:

The program significantly enhances data integrity and analysis in the presence of disconnected EEG electrodes. By predicting the values of these electrodes, researchers can maintain the quality of data and derive more accurate insights. Experiment outcomes are improved, enabling reliable and robust conclusions. Moreover, the program offers a cost-effective solution for non-professional grade EEG devices, expanding research and clinical possibilities.


The development of a program capable of predicting disconnected EEG electrode values using bidirectional LSTM networks presents a breakthrough in EEG data analysis. Overcoming challenges posed by disconnected electrodes, this program provides valuable insights and enhances experiment outcomes for our client. The case study has highlighted the development process, the functionality and features of the program, and the potential benefits and impact on EEG data analysis. Future advancements and applications hold the potential to further revolutionize EEG research and clinical practices.


Related blogs