Tyto Insights for Wheeze Detection

The Tyto Insights for Wheeze Detection is an AI-driven software system that analyzes lung sound recordings from a compatible Tyto Stethoscope to assist in identifying suspected wheezing in adults and pediatric patients (2 years and older). Its core science involves an artificial intelligence algorithm, trained on real-world acoustic data, which processes these sounds to provide a binary output (wheeze suspected/not suspected); validation studies showed a sensitivity of 0.54, specificity of 0.98, and non-inferiority to a rule-based predicate device with a higher AUC.

The Tyto Insights for Wheeze Detection is an over-the-counter, artificial intelligence-enabled software system designed for evaluating lung sounds in adults and children aged two years and older. It functions by automatically analyzing acoustic signals of the lungs, as recorded by a compatible Tyto Stethoscope, to identify instances where wheezing is suspected. The system comprises an Application Server, an AI-driven Algorithm Server that processes audio files and outputs a wheeze detection result (positive, negative, or unable to analyze), and a Web Server that graphically displays this result to both patients and clinicians. It is explicitly intended only for wheeze detection and not for identifying other lung sounds, with the results requiring interpretation by a licensed healthcare professional in conjunction with other relevant patient information. The device's performance was evaluated retrospectively using a dataset of 371 lung sound recordings from 359 patients (ages 2 and older), with ground truth established by a majority vote of three blinded pulmonologists. For stand-alone accuracy, this evaluation yielded a sensitivity of 0.54 and a specificity of 0.98 for wheeze detection. In a comparative analysis against its predicate device, which uses a rule-based algorithm, the Tyto Insights for Wheeze Detection met its primary endpoint of non-inferiority and demonstrated an Area Under the Curve (AUC) of 0.9654, numerically higher than the predicate's AUC of 0.9085, for detecting wheezes. The AI algorithm was trained using real-world recordings from the compatible Tyto Stethoscope, and the system processes recordings in segments up to 12 seconds, not processing signals shorter than 6 seconds.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. Given the sensitivity of 0.54, what is the potential clinical impact of false negatives, especially in acute respiratory distress scenarios, and how should users be guided to mitigate this risk?; How does the AI algorithm differentiate wheezes from other adventitious lung sounds (e.g., rhonchi, crackles) or common background noises, and what were the specific features or datasets used for training this differentiation?; What is the variability in performance (sensitivity/specificity) across different age sub-groups within the pediatric population (e.g., 2-5 years vs. 6-12 years vs. adolescents) and in patients with varying severities of wheezing or comorbidities other than COPD or Asthma?; How does the device's performance change in real-world scenarios with varying levels of user proficiency in stethoscope placement and recording quality, particularly for lay users in a home environment?; Considering the Positive Predictive Value of 0.72, what further diagnostic steps or clinical assessments are typically recommended following a 'wheeze suspected' result from the device to confirm the diagnosis and determine the underlying etiology?