Frequently Asked Questions

CoughMonitor Suite (CMS) is an end-to-end platform for objective, continuous cough monitoring designed specifically for clinical trials and life sciences. It consists of a wrist-worn device, a companion mobile app, a secure cloud backend, and a web-based researcher dashboard. The system detects and time-stamps cough events passively and continuously without recording any audio.

CMS is developed by , global market leader in AI powered cough monitoring.

CMS is purpose-built for pharmaceutical sponsors, contract research organizations (CROs), and academic investigators running clinical trials in which cough is an endpoint, an operational signal, or a disease-activity measure. Typical use cases include drug development, respiratory disease trials (chronic cough, COPD, IPF, asthma, bronchiectasis), and studies where cough provides supporting clinical context.

• Wearable: a smartwatch running on-device acoustic AI that detects cough events in real time without recording audio.

• Companion app: an iOS/Android application that syncs cough timestamps from the watch via Bluetooth and collects validated patient-reported outcomes (PROs), including VAS cough severity, LCQ, and custom study-specific questionnaires.

• Cloud backend: a secure, GDPR- and HIPAA-aligned infrastructure that stores derived event data (timestamps, counts, coverage metadata) - never raw audio.

• Researcher dashboard: a web-based interface providing real-time adherence monitoring, data quality verification, and data download for study teams.

Cough detection is powered by an AI that is embedded in the smartwatch firmware which works in two phases. In a first phase explosive sudden acoustic events with the temporal envelope characteristic of cough get flagged. In a second phase, a cough-specific classifier applies discriminative learned features - spectral structure, transient characteristics, timing patterns - to confirm the event as a cough and reject confounders such as speech, laughter, throat clearing, and environmental noise.

All processing occurs at the edge, on the watch, in real time. The models do not require audio to leave the device. Only text-format timestamps are transmitted to the companion app and cloud backend. This architecture is a fundamental design choice and the basis of Hyfe's privacy preservation guarantee.

To learn more about how Hyfe’s technology works, .

No. Audio is processed transiently on the device to produce a binary classification (cough / not cough), and then discarded. No raw audio is stored on the watch, transmitted to the phone, or held on Hyfe's servers at any point in the standard operating mode.

This is categorically different from ambulatory audio recording systems, where raw audio is retained and reviewed - a practice that creates persistent identifiable data, increases IRB scrutiny, requires secure storage infrastructure, and presents privacy risks that Hyfe's approach eliminates by design.

In limited, separately governed research configurations (e.g., algorithm validation studies), optional audio capture can be enabled with explicit participant consent, a separate consent process, and a dedicated technical stack. This is not part of standard CMS trial deployment.