Dispersed Clinical Validation of the Hyfe Cough Monitoring System
Dispersed Clinical Validation of the Hyfe Cough Monitoring System
This study will test the accuracy of the Hyfe Cough Monitoring System. This system uses a smartwatch to record the number and pattern of coughs (like a "Fitbit for cough").
The investigators will compare the coughs counted by the Hyfe device with manual cough counts from environmental sounds recordings in the same period.
This is a dispersed observational study and will include individuals with a variety of cough related conditions including but not limited to refractory or unexplained cough, Bronchiectasis, COPD, COVID-19, non-tuberculous mycobacterial infection, tuberculosis, or upper respiratory tract infection.
The Hyfe Cough Monitoring System is investigational, which means that it has not been approved by the Food and Drug Administration (FDA). This study is funded by Hyfe, Inc. and executed under the direction of Dr. Peter Small (the Investigator). The study will run until thirty (30) participants have completed the study
The Primary objective is to assess the agreement in cough counting between the Hyfe Cough Monitoring System and human listeners on an hour by hour basis when used by individuals with problematic cough under common living conditions using a Bland-Altman analysis.
The Hyfe Watch, along with the companion app, is part of the Hyfe Cough Monitoring System. It does not compromise the users provacy as it only retains the time of day associated with each cough. When the system detects a short explosive sound, it runs a brief "snippet" of that sound through a cough detection algorithm on the watch. If the algorithm deems the sound to be a cough, the device records the exact time that the cough occurred, and sends this time stamp to the companion app. The companion app does not access any personal information on the phone and is only used to "collect" the times at which a participant coughed from the Hyfe Watch and send them digitally to the researcher.
At the end of the 24 hours monitoring period envisiones, participants will be contacted by video call and asked to download this information from the companion app and send this to the Investigator. The use of the Hyfe Cough Monitoring System will not lead to additional medical interventions or changes to medical care.
In contrast, the sound recording watch will record all ambient sounds during a 24 hour period. This is necessary for the investigator to get an independent count of a participant´s coughing. As such, the sound recording watch will record conversations and those in the participant´s environment during this 24 hour period. It is important to let others in the environment know that sound is being recorded. The recordings will not be directly linked with the participant´s identity and will only be listened to by researchers who are trained in how to ensure the privacy of this data.
The data collected from the sound recording watch will NOT be transferred through the internet while participants are wearing it. Audio recordings will be saved on the actual device. Investigators will download the audio recordings onto a computer after the return the devices. Recordings will then be fully deleted.
Criteria for success:
The goal of this study is to demonstrate the performance of the Hyfe Cough Monitoring System as a device that continually and unobtrusively monitors cough. As has been agreed upon with the FDA, There is no pre-specified performance goal for the HCMS, this information will be reported in the labeling material. Therefore, the success of this study will be judged based on demonstrating adequate precision of our endpoints as quantified by their margin of error. Thus, the study will be considered satisfactory if the following criteria are satisfied:
Bland-Altman analysis:
These 95% Bland-Altman Limits of Agreement (LOA) criteria correspond to the extreme acceptable differences between Hyfe's cough-second counts and those of human annotators; by requiring these limits to have magnitude at most 5, it is expected that most of the differences to have magnitude less than 5. Note that the accompanying criterion for success for the bias is that its confidence interval have endpoints of magnitude at most 0.5, so it is expected that the average difference to be close to 0.
Inclusion Criteria:
Exclusion Criteria: