Search Results

The Ceribell Delirium Monitor System is intended to analyze features associated with diffuse slowing electroencephalogram (EEG) patterns that may be indicative of delirium. The Ceribell Delirium Monitor software is intended to aid in the screening and monitoring of delirium with clinical assessments in adult patients in critical care settings within hospitals.

The Ceribell Delirium Monitor System analyzes discrete segments of EEG to notify clinicians when EEG patterns associated with delirium are detected while monitoring the patient. Changes in patient condition that are detected by the device should be verified before commencing any interventions.

Ceribell's Delirium Monitor System is a novel device that uses EEG to aid in the detection of delirium. The device is comprised of the Delirium Monitor software as well as Ceribell's previously cleared EEG acquisition system (K191301) and EEG headband (K232052) for recording and processing the EEG.

Once the EEG is recorded and processed the Delirium Monitor software performs the following steps:

1. Extraction of EEG features relevant to delirium assessment in 30-minute segments
2. Analysis of the extracted features of each segment with a machine-learning based delirium assessment model
3. Providing delirium-positive or delirium-negative assessment output to the intended users after the first 30 minutes of recording, including audible and visible notifications on the Ceribell Pocket EEG Device in the event of delirium-positive findings
4. Thereafter, providing delirium-positive or delirium-negative assessment output to the intended users once every 15 minutes, generating a notification when the assessment transitions from negative to positive.

Once generated, the output of the Delirium Monitor will be interpreted by the clinician alongside other relevant information to determine whether a diagnosis of delirium is appropriate.

This document describes the acceptance criteria and the study proving the Ceribell Delirium Monitor System meets these criteria, based on the provided FDA 510(k) Clearance Letter.

Acceptance Criteria and Reported Device Performance

Study Details

1. Sample Size used for the Test Set and Data Provenance:

   • Sample Size: 225 adult patients (22 years or older).
   • Data Provenance: Retrospective analysis of EEG data collected from patients in a critical care environment. The country of origin for the data is not specified in the provided text.

2. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of those Experts:

   • Number of Experts: The text states, "Clinical assessment of the patient (according to the DSM-5 criteria for delirium diagnosis) was also conducted up to three times a day by a qualified clinician." It does not specify the exact number of clinicians involved or their precise qualifications beyond "qualified clinician," though it implies expertise in delirium assessment.

3. Adjudication Method for the Test Set:

   • The provided text does not explicitly detail an adjudication method beyond "clinical assessment ... by a qualified clinician." It does not mention committee review, multiple independent clinicians followed by arbitration (e.g., 2+1, 3+1), or other formal adjudication processes for establishing ground truth from disagreements. It implies a single "qualified clinician" established the ground truth for each assessment.

4. Multi Reader Multi Case (MRMC) Comparative Effectiveness Study:

   • No such study was explicitly reported. The provided text focuses on the standalone performance of the algorithm against clinical ground truth. There is no mention of human readers improving with AI assistance, nor any reported effect size for such improvement.

5. Standalone Performance:

   • Yes, a standalone (algorithm only without human-in-the-loop) performance study was done. The performance metrics (Sensitivity, Specificity, AUROC, PPV, NPV, Cohen's Kappa) reported in Table 2 and Table 3 directly represent the algorithm's performance against the established clinical ground truth without human intervention in the device's output interpretation.

6. Type of Ground Truth Used:

   • Expert Consensus (Clinical Diagnosis): The ground truth was established by "clinical assessment of the patient (according to the DSM-5 criteria for delirium diagnosis) ... by a qualified clinician." This indicates expert clinical diagnosis based on established medical criteria as the reference standard.

7. Sample Size for the Training Set:

   • The provided document does not specify the sample size for the training set. It mentions that the "recorded EEG data and corresponding clinical delirium diagnosis were analyzed retrospectively to determine the performance of the Delirium Monitor," and later refers to "the validation dataset." This implies a separate training phase but does not provide its details. The PCCP mentions "expansion of training data and optimization of the algorithm," further suggesting a training set was used but its size is not disclosed here.

8. How the Ground Truth for the Training Set Was Established:

   • The document does not explicitly state how the ground truth for the training set was established. However, given the validation set's ground truth was established by "clinical assessment ... by a qualified clinician" according to DSM-5 criteria, it is highly probable that a similar methodology was used for the training data. The PCCP notes "Ceribell's data management and algorithm development practices," which would likely encompass ground truth establishment for training data.

Ask a specific question about this device