Machine learning based seizure classification and digital biosignal analysis of ECT seizures DOI Creative Commons
Max Kayser, René Hurlemann, Alexandra Philipsen

и другие.

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Фев. 21, 2025

While artificial intelligence has received considerable attention in various medical fields, its application the field of electroconvulsive therapy (ECT) remains rather limited. With advent digital seizure collection systems, development novel ECT quality metrics and treatment guidance systems particular will require cutting-edge analysis. Using offer more analytical degrees freedom could play a key role enhancing precision currently available procedures. To this end, we developed first machine learning (ML) framework that can classify ictal non-ictal EEG segments, accurately identifying endpoints—a critical step deriving parameters—and computing these at least as reliable existing precomputed scores. The ML model retained study effectively discriminated from segments with 89% accuracy, precision, sensitivity. reproduced parameters showed correlations up to ϱ = 0.99 (p < 0.01) pre-calculated values stimulation device did not significantly differ reference values. Mean duration differences were 0.23 ± 15.59 s compared expert rater 0.28 16.19 device. highlights potential integrating into emphasizes highly sensitive detection method reliably determining subsequent indices, paving way for individualized strategies approaches determine quality.

Язык: Английский

Machine learning based seizure classification and digital biosignal analysis of ECT seizures DOI Creative Commons
Max Kayser, René Hurlemann, Alexandra Philipsen

и другие.

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Фев. 21, 2025

While artificial intelligence has received considerable attention in various medical fields, its application the field of electroconvulsive therapy (ECT) remains rather limited. With advent digital seizure collection systems, development novel ECT quality metrics and treatment guidance systems particular will require cutting-edge analysis. Using offer more analytical degrees freedom could play a key role enhancing precision currently available procedures. To this end, we developed first machine learning (ML) framework that can classify ictal non-ictal EEG segments, accurately identifying endpoints—a critical step deriving parameters—and computing these at least as reliable existing precomputed scores. The ML model retained study effectively discriminated from segments with 89% accuracy, precision, sensitivity. reproduced parameters showed correlations up to ϱ = 0.99 (p < 0.01) pre-calculated values stimulation device did not significantly differ reference values. Mean duration differences were 0.23 ± 15.59 s compared expert rater 0.28 16.19 device. highlights potential integrating into emphasizes highly sensitive detection method reliably determining subsequent indices, paving way for individualized strategies approaches determine quality.

Язык: Английский

Процитировано

0