21.03.2024

In the future, artificial intelligence will monitor patients with advanced heart failure telemedically by the sound of their voice. This will make it possible to detect and treat deterioration at an early stage. The Berlin start-up Noah Labs UG has developed the AI required for this.

Dr. Leonhard Riehle

In an interview with MEDICA-tradefair.com, Dr. Leonhard Riehle explains how telemedical monitoring of heart failure can work with the help of AI.

Noah Labs specializes in the voice-based diagnosis of heart disease. Dr. Riehle, why did you decide to focus on the voice in combination with AI?

Dr. Leonhard Riehle: The human voice holds immense potential. Everyone has a voice and uses it every day. We are only just beginning to understand the connections between the voice and health. Today, we can already predict the risk of disease with astonishing accuracy based on the voice. This development was largely made possible by the rapid progress of AI.

As heart disease affects almost everyone in their lifetime and remains the most common cause of death in Germany, the combination of voice-based diagnostics and cardiovascular disease has great potential. By using the voice, the technology is easily accessible, usable and cost-effective for almost everyone affected.

For those treating patients, the voice has the advantage that it may provide earlier indications of deteriorating health than conventional methods. In addition, closer monitoring is also possible remotely, which noticeably increases the quality of care. The ultimate aim is to reduce hospitalizations and deaths for those affected, while relieving the burden on doctors and the healthcare system as a whole.

Products and exhibitors around this topic

Discover more interesting products and exhibitors in the database of MEDICA:

Image: Woman holding a cell phone in her hand in front of her face and speaking into it

For the analysis, the AI breaks down the patient's voice into many individual "features" such as frequency, pitch and volume.

What are the most important parameters that make it possible to monitor advanced heart failure by the sound of the voice?

Riehle: Chronic heart failure leads to increased water retention in the body. Most people are familiar with the classic swollen legs in older people, a common sign of heart failure. However, this water retention also occurs in the vocal tract and leads to subtle changes in the voice that cannot be detected by the human ear alone.

Our AI breaks down the patient's voice into many individual components, known as "features", such as frequency, pitch, volume, etc. The AI compares these features with each other. Then the AI compares these values with previous recordings. In this way, the smallest changes can be detected at an early stage and an impending deterioration due to increased water retention can be counteracted.

How does the AI have to be trained for this?

Riehle: The AI is trained on the basis of existing patient data. The issue of data protection is very important; all voice data is only ever collected with the consent of those affected. A large number of data sets are required, which are recorded under controlled clinical conditions. Each voice sample is annotated with further medical data and provides precise information about the state of health.

In order to collect this data in such high quality, we have already conducted a large number of studies with leading university hospitals in Europe, such as the German Heart Center of the Charité. Further studies will start shortly in which our voice analysis will support medical professionals in the care of heart failure patients in real time.

The aim here is also to prove the effectiveness in terms of avoiding hospital stays and reducing mortality.

What is necessary for qualitative monitoring?

Riehle: In order to achieve the highest possible quality of monitoring, several prerequisites are necessary. Firstly, the AI must be trained on the basis of high-quality data. A large number of training data sets is essential. The quality of the AI must also be examined in clinical studies and tested with regard to efficacy and safety.

However, the technology is very easy for patients to use. A standard smartphone is all that is needed to carry out the daily voice recording.

What are the next steps for Noah Labs?

How does telemedical voice monitoring with AI work?

Image: Infographic on the process of an AI analysis of the voice in heart failure in key points with corresponding icons; Copyright: beta-web GmbH/Riehle"

If there are signs of a deterioration in health, the medical staff treating the patient are automatically informed and can initiate appropriate medical measures immediately.

Riehle: The next step in the current ongoing study is the completion of patient recruitment at the German Heart Center of the Charité and the Mayo Clinic in the USA. Further studies are already planned and will provide more data on the effectiveness of voice-based monitoring. Subsequently, the aim is to obtain approval for the solution as software and as a medical device in the near future in order to be able to bring the solution to the medical care sector.

In addition to heart failure, the voice also has enormous potential to improve the treatment of chronic illnesses in a number of other diseases. Other cardiovascular diseases, diabetes mellitus and chronic lung diseases show great potential for the use of the technology. Expansion into other countries is also planned in the future.