Search Results

The Sunrise SDDA device is a non-invasive home care aid in the evaluation of obstructive sleep apnea (OSA) in patients 18 years and older with suspicions of sleep breathing disorders.

The Sunrise SDDA device consists of a Sunrise sensor and a cloud-based software device that analyzes data from the sensor when placed on the patient's mandible. The device also includes a mobile application to record patient's responses to questions about their sleep quality and transfer sensor data to the cloud. By analyzing patient's mandibular movements, the device also detects obstructive respiratory disturbances, identifies sleep states, notifies about the Obstructive Sleep Apnea (OSA) severity in a categorical format (non-OSA, mild-OSA, moderate-OSA, severe-OSA), generates sleep structure information (namely, total sleep time, sleep onset latency, wake after sleep onset, sleep efficiency, arousal index) and head position discrete states. Data collected by the device is integrated in a report for further interpretation and diagnosis by the healthcare provider.

Here's a breakdown of the acceptance criteria and study information for the Sunrise Sleep Disorder Diagnostic Aid (SDDA), based on the provided text:

Acceptance Criteria and Reported Device Performance

• Sensitivity: (b)(4)
• Specificity: (b)(4)
  *For ORDI cut-offs: ORDI> (b)(4), ORDI> (b)(4), and ORDI> (b)(4) events/h, respectively. * (Specific values for each cutoff are redacted as (b)(4)).

Third Study (Belgium):

• Sensitivity: (b)(4)
• Specificity: (b)(4)
  *For ORDI cut-offs: ORDI>= (b)(4), ORDI>= (b)(4), and ORDI>= (b)(4) events/h, respectively. * (Specific values for each cutoff are redacted as (b)(4)). |
  | Sleep Structure Parameters (TST, SOL, WASO, SE, ArI) Including Total Sleep Time (TST), Sleep Onset Latency (SOL), Wake After Sleep Onset (WASO), Sleep Efficiency (SE), Arousal Index (ArI) | The clinical data must demonstrate output consistency and compare device performance with a clinical comparator device (polysomnography). Performance should be quantified by Root Mean Square Error (RMSE) and confidence intervals (CIs). | First Study (Belgium - Retrospective):
• RMSE: (b)(4) (CI (b)(4)) for TST, SOL, WASO, SE, and ArI respectively. (Specific values for each are redacted as (b)(4)).

Second Study (France - Prospective):

• RMSE: (b)(4) (CI (b)(4)) for TST, SOL, WASO, SE, and ArI respectively. (Specific values for each are redacted as (b)(4)).

Third Study (Belgium - Retrospective):

• RMSE: (b)(4) (CI (b)(4)) for TST, SOL, WASO, SE, and ArI respectively. (Specific values for each are redacted as (b)(4)). |
  | Biocompatibility | Demonstrate that patient-contacting components are biocompatible. | Test articles (skin adhesive and film dressing) found non-cytotoxic, non-sensitizing, and non-irritating per ISO 10993-5 and ISO 10993-10. |
  | Electromagnetic Compatibility & Electrical Safety | Performance data must be provided to demonstrate EMC and electrical, mechanical, and thermal safety. | IEC 60601-1 and IEC 60601-1-2 testing performed; results support electrical safety and electromagnetic compatibility. |
  | Software Validation | Appropriate documentation to support validation for a Moderate Level of Concern, including algorithms, hardware characteristics, and mitigations for subsystem failures. Cybersecurity measures also addressed. | "Appropriate documentation" provided per FDA's 2005 (Software) and 2014 (Cybersecurity) guidance documents, including workflow, handling of errors, and algorithm development steps. |
  | Human Factors/Usability | Usability engineering testing in accordance with IEC 62366-1:2015 should demonstrate that safety-related tasks can be successfully performed. | Formative usability testing conducted in Belgium with adult users (tech-savvy & non-tech-savvy). Majority of participants completed all tasks correctly. Outcome assessed as satisfactory, providing "adequate assurance that all tasks linked to a safety mitigation could be successfully performed." No critical tasks were identified that could result in serious harm if performed incorrectly. |
  | Packaging and Shelf Life | Packaging and labeling should withstand anticipated shipping conditions and preserve functionality. Shelf-life determined and supported. | Drop testing, resistance to rain/humidity, and label integrity evaluations demonstrated appropriate protection. Shelf-life of 2 years determined based on adhesive shelf-life. |

Study Details

The sponsor provided three clinical studies to support the safety and effectiveness of the Sunrise SDDA device.

1. First Clinical Study (Retrospective)

• Sample Size: Not explicitly stated for this particular study, but described as "patients."
• Data Provenance: Belgium, retrospective.
• Number of Experts for Ground Truth: One experienced sleep technician.
• Qualifications of Experts: "Experienced sleep technician."
• Adjudication Method for Test Set: None explicitly mentioned as a multi-expert adjudication process. The PSG data was visually scored by a single experienced sleep technician.
• MRMC Comparative Effectiveness Study: No. This study focused on algorithm performance against PSG.
• Standalone Performance: Yes, the Sunrise Machine Learning algorithms analyzed sensor data to evaluate the device performance for sleep structure parameters compared to in-lab PSG.
• Type of Ground Truth: Expert-scored Polysomnography (PSG) by an experienced sleep technician,
  following 2012 AASM recommendations, and blinded to the study protocol.
• Sample Size for Training Set: Not explicitly stated, however, the text mentions that "the same datasets were used for both optimizing diagnostic thresholds (training) and performance evaluation (validation)," suggesting this study may have contributed to or been part of the training data.
• How Ground Truth for Training Set was Established: PSG data visually scored by an experienced sleep technician according to 2012 AASM recommendations.

2. Second Clinical Study (Prospective)

• Sample Size: Not explicitly stated, described as "patients."
• Data Provenance: France, prospective.
• Number of Experts for Ground Truth: Not explicitly stated beyond "experienced sleep technicians."
• Qualifications of Experts: "Experienced sleep technicians from two different sleep centers (Université Grenoble Alpes, Grenoble, France and Imperial College London, London, United Kingdom)."
• Adjudication Method for Test Set: Not explicitly stated as a formal adjudication protocol (e.g., 2+1), but PSG data was scored by "experienced sleep technicians from two different sleep centers," suggesting independent scoring, though not necessarily an adjudication to resolve discrepancies.
• MRMC Comparative Effectiveness Study: No. This study focused on algorithm performance against PSG.
• Standalone Performance: Yes, the device's performance for all output parameters (OSA severity and sleep structure) was evaluated compared to ambulatory at-home PSG.
• Type of Ground Truth: Expert-scored Polysomnography (PSG) by experienced sleep technicians from two different sleep centers, following 2012 AASM recommendations.
• Sample Size for Training Set: Not mentioned as contributing to the training set. This was an independent prospective study.
• How Ground Truth for Training Set was Established: Not applicable; this study was for validation.

3. Third Clinical Study (Retrospective)

• Sample Size: Not explicitly stated, described as "patients."
• Data Provenance: Belgium, retrospective.
• Number of Experts for Ground Truth: One experienced sleep technician.
• Qualifications of Experts: "Experienced sleep technician."
• Adjudication Method for Test Set: None explicitly mentioned as a multi-expert adjudication process. The PSG data was visually scored by a single experienced sleep technician.
• MRMC Comparative Effectiveness Study: No. This study focused on algorithm performance against PSG.
• Standalone Performance: Yes, the Sunrise SDDA algorithms analyzed sensor data to evaluate the performance of the device compared to in-lab PSG.
• Type of Ground Truth: Expert-scored Polysomnography (PSG) by an experienced sleep technician,
  following 2012 AASM recommendations, and blinded to the study protocol.
• Sample Size for Training Set: Not explicitly stated. The study is described as "independent clinical study with similar design as the first one," but doesn't mention its role in training.
• How Ground Truth for Training Set was Established: Not applicable; this study was for validation.

Summary of Training and Validation Data Distinction:

• The First Clinical Study was noted to have used "the same datasets... for both optimizing diagnostic thresholds (training) and performance evaluation (validation)," which was deemed insufficient on its own for demonstrating reasonable assurance of safety and effectiveness.
• The Second and Third Clinical Studies appear to serve as independent validation studies, utilizing similar methodologies but without the noted confounding factor of using the same data for training and testing. The second study used the final Sunrise sensor and was prospective, while the third was retrospective like the first.

Key takeaway on training data: While specific training set sizes are not provided, the first study implicitly indicates that some of its data (or data from a similar source) was used for "optimizing diagnostic thresholds (training)." The methods for establishing ground truth for any training data would align with the method used for the first study's ground truth, i.e., expert-scored PSG.

Ask a specific question about this device