Creation of a clinically annotated, multimodal biosignal database of endoscopy sedation

Creation of a clinically annotated, multimodal biosignal database of endoscopy sedation - Endoscopy biosignal database

Status

Recruiting

Phases

Unknown

Study type

Observational

Source

DRKS

Registry ID

DRKS00030290

Enrollment

1272

Registered

2024-01-02

Start date

2023-07-27

Completion date

Unknown

Last updated

2025-04-07

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

sedation in endoscopy

Interventions

Group 1: Intervention - real-time registration of biosignals and clinical events: - "classical" vital parameters (ECG, pulse oximetry, non-invasive blood pressure) - frontotemporal 2-channel EEG -

Eligibility

Sex/Gender

All

Age

18 Years to No maximum

Inclusion criteria

Inclusion criteria: Persons of full age who are capable of giving consent and who have undergone elective endoscopy under sedation and anticipated endoscopy duration of at least 20 minutes are included.

Exclusion criteria

Exclusion criteria: A known structural brain disease, cerebral neoplasia or cerebral metastatic malignancy, a known seizure disorder or impaired consciousness with Glagow Coma Scale < 15 are exclusion criteria. Furthermore, patients with profound hearing loss or allergies to the sedatives administered are excluded.

Design outcomes

Primary

Measure	Time frame
Creation of a clinically annotated, multimodal biosignal database of sedation in endoscopy [without endpoint]. Surrogate objective for the primary objective: Development of a valid, reliable and robust data-driven artificial intelligence model for EEG-based prediction of consciousness status and sedation depth [surrogate endpoint: AUC of consciousness status classification and PK score for clinical sedation depth classification].	—

Measure

Time frame

Creation of a clinically annotated, multimodal biosignal database of sedation in endoscopy [without endpoint]. Surrogate objective for the primary objective: Development of a valid, reliable and robust data-driven artificial intelligence model for EEG-based prediction of consciousness status and sedation depth [surrogate endpoint: AUC of consciousness status classification and PK score for clinical sedation depth classification].

—

Secondary

Measure	Time frame
- Detection and prediction of safety-relevant clinical events, such as (Serious) Adverse Events and too deep sedation [Endpoint: AUCs to predict AEs and SAEs. Classification accuracy for detecting burst suppression patterns as signs of deep sedation]. - Optimisation of known pEEG parameters - Improved predictions of fluctuating depth of sedation - Improved differentiation of shallow sedation depths [Endpoint: AUC/PK values of classification of state of consciousness and clinical depth of sedation]. - Strengthening the robustness of pEEG parameters [Endpoint: AUC/PK values of the classification of consciousness status and clinical sedation depth as a function of the epochal artefact fraction]. - Further development of modern deep learning methods (with representation learning) and comparison with classical ML approaches (without representation learning) [Endpoint: AUC of classification of consciousness status and PK value for classification of clinical sedation depth].	—

Measure

Time frame

- Detection and prediction of safety-relevant clinical events, such as (Serious) Adverse Events and too deep sedation [Endpoint: AUCs to predict AEs and SAEs. Classification accuracy for detecting burst suppression patterns as signs of deep sedation]. - Optimisation of known pEEG parameters - Improved predictions of fluctuating depth of sedation - Improved differentiation of shallow sedation depths [Endpoint: AUC/PK values of classification of state of consciousness and clinical depth of sedation]. - Strengthening the robustness of pEEG parameters [Endpoint: AUC/PK values of the classification of consciousness status and clinical sedation depth as a function of the epochal artefact fraction]. - Further development of modern deep learning methods (with representation learning) and comparison with classical ML approaches (without representation learning) [Endpoint: AUC of classification of consciousness status and PK value for classification of clinical sedation depth].

—

Countries

Germany

Contacts

Public ContactJakob Garbe

Universitätsklinikum Halle (Saale)Klinik für Innere Medizin I

jakob.garbe@uk-halle.de0345 557 2928

Outcome results

None listed

Source: DRKS (via WHO ICTRP) · Data processed: Feb 4, 2026