Abstract
Computational cost is an important consideration for memory encoding prediction models that use data from dozens of implanted electrodes. We propose a method to reduce computational expense by selecting a subset of all the electrodes to build the prediction model. The electrodes were selected based on their likelihood of measuring brain activity useful for predicting memory encoding better than chance (in terms of AUC). A logistic regression prediction model was built using spectral features of intracranial electroencephalography (iEEG) from the selected electrodes. We demonstrate our method on iEEG data from 37 human subjects performing free recall verbal short-term memory tasks. The method achieves a 36.3% reduction in the number of electrodes used for prediction, resulting in a 64.9% reduction in inference computation time with just a 0.3% loss in prediction performance compared to the case when all electrodes were used. The electrodes selected using our method provided improved prediction performance compared to those electrodes that were not selected in 31 out of 37 patients. Building upon this observation, we also developed a method to identify the subjects for whom the proposed electrode selection method would be beneficial.
| Original language | English (US) |
|---|---|
| Title of host publication | 9th International IEEE EMBS Conference on Neural Engineering, NER 2019 |
| Publisher | IEEE Computer Society |
| Pages | 323-327 |
| Number of pages | 5 |
| ISBN (Electronic) | 9781538679210 |
| DOIs | |
| State | Published - May 16 2019 |
| Event | 9th International IEEE EMBS Conference on Neural Engineering, NER 2019 - San Francisco, United States Duration: Mar 20 2019 → Mar 23 2019 |
Publication series
| Name | International IEEE/EMBS Conference on Neural Engineering, NER |
|---|---|
| Volume | 2019-March |
| ISSN (Print) | 1948-3546 |
| ISSN (Electronic) | 1948-3554 |
Conference
| Conference | 9th International IEEE EMBS Conference on Neural Engineering, NER 2019 |
|---|---|
| Country | United States |
| City | San Francisco |
| Period | 3/20/19 → 3/23/19 |
Fingerprint
ASJC Scopus subject areas
- Artificial Intelligence
- Mechanical Engineering
Cite this
A Computationally Efficient Model for Predicting Successful Memory Encoding Using Machine-Learning-based EEG Channel Selection. / Saboo, Krishnakant V.; Varatharajah, Yogatheesan; Berry, Brent M.; Sperling, Michael R.; Gorniak, Richard; Davis, Kathryn A.; Jobst, Barbara C.; Gross, Robert E.; Lega, Bradley; Sheth, Sameer A.; Kahana, Michael J.; Kucewicz, Michal T.; Worrell, Gregory A.; Iyer, Ravishankar K.
9th International IEEE EMBS Conference on Neural Engineering, NER 2019. IEEE Computer Society, 2019. p. 323-327 8717057 (International IEEE/EMBS Conference on Neural Engineering, NER; Vol. 2019-March).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - A Computationally Efficient Model for Predicting Successful Memory Encoding Using Machine-Learning-based EEG Channel Selection
AU - Saboo, Krishnakant V.
AU - Varatharajah, Yogatheesan
AU - Berry, Brent M.
AU - Sperling, Michael R.
AU - Gorniak, Richard
AU - Davis, Kathryn A.
AU - Jobst, Barbara C.
AU - Gross, Robert E.
AU - Lega, Bradley
AU - Sheth, Sameer A.
AU - Kahana, Michael J.
AU - Kucewicz, Michal T.
AU - Worrell, Gregory A.
AU - Iyer, Ravishankar K
PY - 2019/5/16
Y1 - 2019/5/16
N2 - Computational cost is an important consideration for memory encoding prediction models that use data from dozens of implanted electrodes. We propose a method to reduce computational expense by selecting a subset of all the electrodes to build the prediction model. The electrodes were selected based on their likelihood of measuring brain activity useful for predicting memory encoding better than chance (in terms of AUC). A logistic regression prediction model was built using spectral features of intracranial electroencephalography (iEEG) from the selected electrodes. We demonstrate our method on iEEG data from 37 human subjects performing free recall verbal short-term memory tasks. The method achieves a 36.3% reduction in the number of electrodes used for prediction, resulting in a 64.9% reduction in inference computation time with just a 0.3% loss in prediction performance compared to the case when all electrodes were used. The electrodes selected using our method provided improved prediction performance compared to those electrodes that were not selected in 31 out of 37 patients. Building upon this observation, we also developed a method to identify the subjects for whom the proposed electrode selection method would be beneficial.
AB - Computational cost is an important consideration for memory encoding prediction models that use data from dozens of implanted electrodes. We propose a method to reduce computational expense by selecting a subset of all the electrodes to build the prediction model. The electrodes were selected based on their likelihood of measuring brain activity useful for predicting memory encoding better than chance (in terms of AUC). A logistic regression prediction model was built using spectral features of intracranial electroencephalography (iEEG) from the selected electrodes. We demonstrate our method on iEEG data from 37 human subjects performing free recall verbal short-term memory tasks. The method achieves a 36.3% reduction in the number of electrodes used for prediction, resulting in a 64.9% reduction in inference computation time with just a 0.3% loss in prediction performance compared to the case when all electrodes were used. The electrodes selected using our method provided improved prediction performance compared to those electrodes that were not selected in 31 out of 37 patients. Building upon this observation, we also developed a method to identify the subjects for whom the proposed electrode selection method would be beneficial.
UR - http://www.scopus.com/inward/record.url?scp=85066738980&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85066738980&partnerID=8YFLogxK
U2 - 10.1109/NER.2019.8717057
DO - 10.1109/NER.2019.8717057
M3 - Conference contribution
T3 - International IEEE/EMBS Conference on Neural Engineering, NER
SP - 323
EP - 327
BT - 9th International IEEE EMBS Conference on Neural Engineering, NER 2019
PB - IEEE Computer Society
ER -