Clinical Validation of the Hyfe Cough Monitoring System
- Conditions
- Cough
- Interventions
- Device: Clinical Validation of the Hyfe Cough Monitoring System
- Registration Number
- NCT05723159
- Lead Sponsor
- Hyfe Inc
- Brief Summary
This is a multicenter observational study and will include individuals with a variety of cough related conditions including but not limited to refractory chronic cough, Chronic Obstructive Pulmonary Disease (COPD) and non-tuberculous mycobacteria.
The primary objective of this study is to assess the overall performance of the Hyfe Cough Monitoring System (HCMS, Hyfe Inc., 2022) when used by individuals with problematic cough, under common living conditions.The monitoring period for outpatients will be 24 hours.
- Detailed Description
Researcher will evaluate the accuracy of the mobile HCMS as a tool to monitor cough. HCMS is a dedicated Android smart watch running specialized software that runs continuously and calculates a patient's cough rate on an hourly basis. After providing informed consent, research subjects will be instructed to wear the HCMS and a second watch serving as an audio recorder. These devices will be kept on the wrist or within 3 feet (\~91 cm) of the mouth, a behavior abetted by keeping the charging station for both watches on the bedside table. Participants will be asked to go about their day as usual while wearing these devices.
The subjects will be instructed to not turn off either watch for the duration of the 24 hours. In addition, participants will be instructed to avoid environments with active coughers and to inform others in their environment that sound is being recorded. Additionally, they will be told to avoid getting the watches wet. Participants will be instructed to charge the watches on a nightstand next to the bed while they sleep using the provided watch chargers. At the end of the 24-hour period, participants will be instructed to turn off and remove the watches. Additionally, participants will be given a printed Hyfe diary to write down the exact time when the device was turned on/off and any times the watch was not being worn. The devices will be returned to the researchers once the recording period is complete.The 24 hours of continuous ambient sounds collected by the audio recording watch and will be listened to by Hyfe technicians trained in cough annotation and data security, and each coughwill be annotated on the audio file using audio-analysis software, according to standard operating procedures developed by Hyfe. These results will be compared on an hourly basis with the coughs detected by HCMS and the performance of the HCMS for identifying cough in home settings calculated in comparison to the human annotated continuous recording reported.
Recruitment & Eligibility
- Status
- ACTIVE_NOT_RECRUITING
- Sex
- All
- Target Recruitment
- 21
- Age 21 years old or above
- Individuals seeking health care, who express concern about their active cough (problematic cough)
- Anticipate that they can collect auditory recording and keep the devices with them continuously for 24 hours.
- Residing in a domestic environment without unusually high and / or persistent background sound levels .
- Willing to wear a watch an audio recorder and keep them at bedside (within 3 ft) from the mouth during the night.
- Inability to accept the privacy policy and terms of use of HCMS due to confidentiality or other concerns.
- Inability to avoid unusually prolonged loud environments for the duration of the 24-hour study period.
- Need to conduct confidential conversations during the 24-hour monitoring period
- Individuals who have had significant change in antitussive therapy in the week preceding study.
Study & Design
- Study Type
- OBSERVATIONAL
- Study Design
- Not specified
- Arm && Interventions
Group Intervention Description OHSU site Clinical Validation of the Hyfe Cough Monitoring System Research subjects will be enrolled at OHSU. Research subjects will be instructed to wear the HCMS and a second watch serving as an audio recorder. These devices will be kept on the wrist or within 3 feet (\~91 cm) of the mouth, a behavior abetted by keeping the charging station for both watches on the bedside table. Participants will be asked to go about their day as usual while wearing these devices. Clínica Universidad Navarra site Clinical Validation of the Hyfe Cough Monitoring System Research subjects will be enrolled at Clínica Universidad Navarra. Research subjects will be instructed to wear the HCMS and a second watch serving as an audio recorder. These devices will be kept on the wrist or within 3 feet (\~91 cm) of the mouth, a behavior abetted by keeping the charging station for both watches on the bedside table. Participants will be asked to go about their day as usual while wearing these devices. Decentralized - US-based Clinical Validation of the Hyfe Cough Monitoring System Research subjects will be enrolled in a decentralized manner ou of the Hyfe North American Clinical Office. Research subjects will be instructed to wear the HCMS and a second watch serving as an audio recorder. These devices will be kept on the wrist or within 3 feet (\~91 cm) of the mouth, a behavior abetted by keeping the charging station for both watches on the bedside table. Participants will be asked to go about their day as usual while wearing these devices.
- Primary Outcome Measures
Name Time Method Hyfe Cough Monitoring System (HCMS) false positivity rate Each participant will be monitored continuously with the audio recorder and the HCMS for 24 hours, and cough events will be tabulated hourly. Each participant will thus contribute 24 data points to the calculation of the co-primary outcome measures. The HCMS timestamps coughs as they occur. Hourly counts of cough events will be compared with ground truth hourly counts determined by trained human annotators, who will listen to audio recordings and timestamp each cough using proprietary labeling software.
The false positive rate is the total number of false positives divided by the total number of monitoring hours.Hyfe Cough Monitoring System (HCMS) accuracy using Pearson correlation coefficient Each participant will be monitored continuously with the audio recorder and the HCMS for 24 hours, and cough events will be tabulated hourly. Each participant will thus contribute 24 data points to the calculation of the primary outcome measures. The HCMS timestamps coughs as they occur. Hourly counts of cough events will be compared with ground truth hourly counts determined by trained human annotators, who will listen to audio recordings and timestamp each cough using proprietary labeling software.
If human annotators and the HCMS agreed perfectly, their paired hourly counts would lie on the line y=x (ground truth on the x-axis, HCMS on the y-axis). Agreement with this ideal line quantifies HCMS performance: the Pearson correlation coefficient must be close to 1
By comparison with ground truth annotations, each HCMS timestamp is either a true positive or a false positive; by the usual formula, HCMS Sensitivity = Number of true positives Total number of coughs where the denominator is determined by ground truth.Hyfe Cough Monitoring System (HCMS) sensitivity of cough detection Each participant will be monitored continuously with the audio recorder and the HCMS for 24 hours, and cough events will be tabulated hourly. Each participant will thus contribute 24 data points to the calculation of the co-primary outcome measures. The HCMS timestamps coughs as they occur. Hourly counts of cough events will be compared with ground truth hourly counts determined by trained human annotators, who will listen to audio recordings and timestamp each cough using proprietary labeling software.
By comparison with ground truth annotations, each HCMS timestamp is either a true positive or a false positive; by the usual formula
HCMS Sensitivity = Number of true positives Total number of coughs
where the denominator is determined by ground truth.Hyfe Cough Monitoring System (HCMS) accuracy using OLS slope/intercept Each participant will be monitored continuously with the audio recorder and the HCMS for 24 hours, and cough events will be tabulated hourly. Each participant will thus contribute 24 data points to the calculation of the co-primary outcome measures The HCMS timestamps coughs as they occur. Hourly counts of cough events will be compared with ground truth hourly counts determined by trained human annotators, who will listen to audio recordings and timestamp each cough using proprietary labeling software.
If human annotators and the HCMS agreed perfectly, their paired hourly counts would lie on the line y=x (ground truth on the x-axis, HCMS on the y-axis). Agreement with this ideal line quantifies HCMS performance: the slope and intercept of the OLS line of best fit must be close to 1 and 0, respectively.
By comparison with ground truth annotations, each HCMS timestamp is either a true positive or a false positive; by the usual formula
HCMS Sensitivity = Number of true positives Total number of coughs
where the denominator is determined by ground truth.
- Secondary Outcome Measures
Name Time Method Dependence of HCMS performance on individual cough rates Each participant will be monitored continuously with the audio recorder and the Hyfe Cough Monitoring System for 24 hours. All of the coughs timestamped by trained human annotators and by the HCMS will be used to calculate the seco It is possible that HCMS performance could depend on the frequency with which different subjects cough. To assess the impact of this variable, subjects will be stratified into tertiles according to their average cough rates; a subject's average cough rate is that individual's total number of coughs divided by the number of hours of monitoring. The performance metrics defined above will be calculated separately for these tertiles and compared.
Differential analysis of HCMS performance for individual subjects Each participant will be monitored continuously with the audio recorder and the Hyfe Cough Monitoring System for 24 hours. All of the coughs timestamped by trained human annotators and by the HCMS will be used to calculate the seco HCMS performance will vary from subject to subject; in particular, the HCMS may fail to recognize extremely unusual coughs or may perform differently with challenging acoustic backgrounds. Trained human annotators face similar issues, leading to disputed annotations that will be resolved by a third expert annotator.
To understand the impact of atypical coughers on HCMS performance, the accuracy metrics defined earlier will be calculated separately for each individual participant. The distributions of these accuracy metrics will be investigated and summarized, and a post hoc analysis of audio from those subjects with exceptionally poor performance will be done to determine the causes of degraded performance.Differential analysis of nighttime versus daytime HCMS performance Each participant will be monitored continuously with the audio recorder and the HCMS for 24 hours. All of the coughs timestamped by trained human annotators and by the Hyfe Cough Monitoring System will be used to calculate the second Given the value of monitoring cough only during the day or only at night, a secondary analysis will compare HCMS performance during these periods; "nighttime" is defined as time spent in bed and "daytime" is defined as time not spent in bed. When going to bed, subjects will note their bedtimes carefully and will place the HCMS and monitor in chargers by the bed; reported bedtimes will be validated against the charging times recorded by the watches. The accuracy metrics defined above will then be calculated separately for subjects' daytime and nighttime monitoring periods.
Trial Locations
- Locations (3)
Oregon Health Science University
🇺🇸Oregon City, Oregon, United States
Hyfe North American Clinical Office
🇺🇸Seattle, Washington, United States
Clínica Universidad de Navarra
🇪🇸Navarro, Spain