Search Results

The AusculThing ACC software is a decision-support SW for the healthcare provider (the user) in the evaluation of patient heart sounds. The ACC is used to record, display, and analyze acoustic signals of the heart recorded by means of an electronic stethoscope. It is intended for use on adult and pediatric patients. The automated analysis will categorize heart sounds as either "abnormal" if any heart murmur of any intensity is identified in any position across the precordium, or "normal" if either no murmurs or benign murmurs are identified. ACC is indicated for use in a setting where auscultation would typically be performed by a healthcare provider. It is not intended as a sole means of diagnosis. The heart sound interpretation offered by the software is only significant when used in conjunction with physician over-read and including all other relevant patient data. The device is intended for Rx use only. The AusculThing ACC shall be used together with Thinklabs One electronic stethoscope.

AusculThing ACC is a decision support SW that collects heart sounds from adult and pediatric patients. The ACC software receives the data using a Thinklabs One electronic stethoscope. The SW is running on a mobile device, where the electronic stethoscope is connected to. The SW guides the user how relevant heart sound recordings should be obtained from different parts of the body. After recording, the ACC analyzes the recordings in conjunction automatically using an AI -based algorithm, which is trained using a proprietary echocardiogram validated high-quality data database. The basic functionality of the ACC SW is to give a user an instant, automated, analysis of the patient under evaluation and differentiate between normal and pathological sounds. For the abnormal heart sounds, the ACC delivers information on suspected murmurs. The ACC software is a SW that allows a user to upload heart sounds/phonocardiogram (PCG) data to the device for analysis and visualization. The AusculThing ACC Mobile App runs on a mobile device. The app permits the electronic recording of heart sound signals via a compatible electronic stethoscope (Thinklabs One). The app also permits visual and acoustic playback of heart sounds in the mobile device. After analysis, results are returned to the user in the App. The Murmur detection algorithm is based on a neural network model that uses heart sounds to detect the presence of pathological heart sounds. The user can utilize the heart sound analysis results and the acoustic and visual representation of the heart sound recordings as decision support data in their decision-making process regarding the presence and type of a heart murmur.

The AusculThing ACC device claims substantial equivalence to the predicate device, eMurmur ID (K181988), for its performance in detecting abnormal heart sounds.

1. Table of Acceptance Criteria and Reported Device Performance

The acceptance criteria for the AusculThing ACC are based on demonstrating non-inferiority to the predicate device, eMurmur ID, in terms of sensitivity, specificity, and accuracy.

The reported performance of the AusculThing ACC (Sensitivity 90.5%, Specificity 96.0%, Accuracy 92.5%) exceeds the performance metrics of the predicate device, eMurmur ID, thereby demonstrating non-inferiority.

2. Sample Size Used for the Test Set and Data Provenance

• Sample Size: The test set comprised 133 patients, from whom a total of 519 heart sound recordings were captured.
  • 84 patients were below 18 years of age.
  • 49 patients were above 18 years of age.
  • 84 patients had a confirmed heart defect.
• Data Provenance: All data was collected in a clinical study conducted in Finland across various hospitals:
  • Children:
    • Kuopio University Hospital (Puijo Hospital)
    • Oulu University Hospital
  • Adults:
    • Hospital district of Helsinki and Uusimaa (Lohja Hospital)
    • Hospital district of Helsinki and Uusimaa (Hyvinkää Hospital)
      The study was conducted in accordance with GCP/ISO14155, indicating a prospective and ethically sound approach to data collection.

3. Number of Experts Used to Establish Ground Truth and Qualifications

• Number of Experts: Not explicitly stated as a number, but the ground truth was established by cardiologists.
• Qualifications: The heart sound recordings were obtained by a cardiologist, and an echocardiogram was conducted by a cardiologist on all patients to establish the golden standard for diagnosis. This implies highly qualified medical professionals experienced in cardiovascular diagnosis.

4. Adjudication Method for the Test Set

The document does not explicitly state an adjudication method (e.g., 2+1, 3+1). It states that an echocardiogram was conducted by a cardiologist on all patients to establish the "golden standard for diagnosis," suggesting that the cardiologist's echocardiogram interpretation served as the definitive ground truth for each case. This implies a single-expert gold standard based on the cardiologist's assessment and the echocardiogram.

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study

• A MRMC comparative effectiveness study was not explicitly conducted or reported in this summary. The comparison is between the standalone performance of the AusculThing ACC algorithm and the reported performance of the predicate device's algorithm, not the improvement of human readers with AI assistance.

6. Standalone (Algorithm Only) Performance

• Yes, a standalone performance study was conducted. The reported sensitivity, specificity, and accuracy values (90.5%, 96.0%, 92.5%) are for the AusculThing ACC algorithm itself, without a human-in-the-loop component for the performance evaluation presented. The device is intended as "decision support SW" and "not intended as a sole means of diagnosis," with interpretation significant "in conjunction with physician over-read," but the reported performance metrics are for the algorithm's direct classification output.

7. Type of Ground Truth Used

• The ground truth used was expert consensus combined with pathology/diagnostic imaging. Specifically, a cardiologist performed an echocardiogram on all patients, which was then used to establish the "golden standard for diagnosis" against which the algorithm's performance was compared.

8. Sample Size for the Training Set

• The document states that the AI-based algorithm was "trained using a proprietary echocardiogram validated high-quality data database." However, the sample size for this training set is not provided in the given text.

9. How the Ground Truth for the Training Set Was Established

• The ground truth for the training set was established using a "proprietary echocardiogram validated high-quality data database." This implies that the training data also had ground truth labels derived from echocardiogram interpretations, likely by cardiologists, similar to how the ground truth for the test set was established. However, specific details about the process for the training set are not provided beyond this general statement.

Ask a specific question about this device