Pre-Symptomatic Detection of Impending Decompensation in Heart Failure through Voice Data (PRE-DETECT-HF)

Heart failure (HF) is a common chronic disease that contains the risk of
imminent volume overload, called decompensation. Symptoms usually occur late
during the course of decompensation, leaving insufficient time to effectively
intervene. Voice-based digital biomarker may detect imminent deterioration
significantly earlier without the requirements of implantation of invasive
devices such as CardioMEMS

The main objective of the study is to optimize the use and user-friendliness of
the voice analysis program, which could potentially help with the early
detection of acute deterioration in heart failure. Hopefully, we can develop a
technique that enables quicker intervention during deterioration. By adjusting
medication more rapidly, the patient may be in the future hospitalized less
frequently, once proven in RCT.

This study will be performed at three sites (Barcelona, Zuyderland,
Maastricht). Patients admitted to hospital with Acutely Decompensated HF (ADHF)
will be included in the study prior to discharge from the hospital. Patients
will be equipped with a smartphone or tablet (if required an additional
external microphone) and the pre-installed Noah Labs patient application (both
provided by Noah Labs to the study sites). The app will include also some
questions regarding symptoms of HF. During admission while still volume
overloaded, patients will perform the first voice recordings. The recording
immediately prior to discharge, i.e. when completely recompensated, will
subsequently be considered the baseline recording. The recording from
decompensation while still volume overloaded until recompensation will be used
to personalize the individual voice pattern, which will be the refence
recording. In their own homes, patients will continue to record structured
voice samples with a combination of pre-defined text samples and a variable
content to avoid boredom daily for about 2 minutes. The variable content will
be related to general worth knowing facts. They will also answer the questions
regarding eventual HF related symptoms (standard care). All voice samples and
the answers to the symptom questions will be transmitted to the Noah Labs
server for analysis and will be compared to the reference sample that was
recorded at the day of discharge of the respective patient.
In addition, investigators at the three clinical sites will be provided access
to the Noah Labs healthcare provider (HCP) platform. If certain thresholds are
exceeded regarding standard care (e.g. increase in shortness of breath,
peripheral oedema), HCP will receive notifications about the findings. It is
also possible for HCP*s to follow daily substantive data.
The study will be 6 months with fixed evaluation at months 1, 3 and 6. At month
1, only a remote contact (by telephone) will take place. After discharge, the
study will be conducted in the outpatient departments of the study sites. The
patients will be seen by a physician knowledgeable of HF that is part of the
study team (investigator). Additional contacts with healthcare providers will
be possible as clinically required.

Recording of structured voice samples with a combination of pre-defined text
samples and a variable content to avoid boredom daily for about 2 minutes. The
variable content will be related to general worth knowing facts. In addition,
standard eHealth in both groups (i.e. questions regarding HF symptoms and body
weight). Based on this monitoring, notifications will be sent to study centres
in case of imminent decompensation. Physician will then intervene with standard
HF treatment to avoid decompensation (mainly diuretics), according to current
HF guidelines.

There is some additional burden for the patient due to the fact that they need
to spend 2-5 minutes daily using the app. Every two weeks, patients will also
need to complete an additional questionnaire with 3-8 questions on various
topics (e.g., expectations, user-friendliness of the app, acceptance of voice
analysis monitoring) via the app. Additionally, 1 extra blood sample will be
collected at each visit.

There is no direct risk associated with using the app. Although an increase in
medication therapy based on deterioration of heart failure has side effects,
such as worsening kidney function, this is generally the case for the treatment
of heart failure and not specific to this study. The associated risks of
worsening heart failure are also estimated to be higher than the side effects
of the medication therapy.