Blood pressure measurement

cosinuss° determines blood pressure directly in the ear canal using a non-invasive, optical measurement technology. The sensor technology in the in-ear sensors is similar to how finger clip pulse oximeters work. A photoplethysmographic signal (PPG) is recorded, from which individual pulse waves are derived. From these, important vital parameters such as: Pulse rate and Oxygen saturation (SpO2) can be derived.

Put simply, the PPG signal is directly related to the blood volume in the vessels in the ear canal, which in turn reflects the blood pressure. If the blood volume increases, the blood pressure also increases. The PPG pulse shape is therefore influenced by the blood pressure, so that changes in blood pressure are directly recognisable in the PPG curve. High blood pressure and arterial vascular stiffness are closely related: High blood pressure makes the arteries stiffer, and stiffer arteries increase systolic blood pressure.¹ Increased blood pressure causes the pressure wave in the arteries to spread faster (pulse wave velocity, PWV). This is reflected in a faster pulse wave and a shorter pulse transit time (PTT). In addition, a higher blood pressure causes a greater systolic amplitude, as the heart exerts more pressure on the blood.² In addition to these visible changes in the PPG curve, there are others. In short, the shape of the PPG wave depends on blood pressure changes that can be traced in the curve (see Fig. 1).

Fig. 1: Schematic amplitude of the pulse curve from a photoplethysmogram (PPG) and its essential features for determining blood pressure.

When measuring PPG, the individual variability of vascular stiffness represents a Challenge dar. The PPG waveforms vary from person to person. cosinuss° therefore breaks down each pulse wave into its individual parts and characterizes its characteristics. This is a very complex task of determining blood pressure with a neural network developed by cosinuss° solved, which works with different layers. The information generated is then translated into blood pressure values. The in Abb. 1 The parameters shown represent a selection of many that cosinuss° uses to algorithmically evaluate the heartbeats to calculate blood pressure. Based on this technology, cosinuss° is able to continuously measure blood pressure in the ear canal.

Measuring blood pressure in the ear canal using optical measuring methods offers decisive advantages – depending on the application.

If blood circulation in the extremities is restricted, as occurs with hypothermia, among other things, reliable blood pressure measurement can no longer take place in these areas. Especially in cold environments, thanks to easy access to the ear, the patient no longer needs to be undressed or a sleeve cut open to free their arm. Aspects that are particularly important in mountain or offshore rescue. Up until now, the decision has often been made as to whether the patient’s blood pressure should be measured, thereby risking hypothermia, or whether this vital parameter should be omitted, meaning that important changes in the patient’s state of health would not be detected in a timely manner.³

In addition, measurements in the ear canal are less susceptible to movement artifacts, as can occur when measuring on the wrist or finger. The dark environment in the ear canal also offers the advantage that external light influences are minimized, which further improves the signal quality.

Another aspect that affects blood circulation in the periphery: When tourniquets are applied to stop heavy bleeding in the extremities, blood pressure measurements no longer work reliably here either. However, this does not affect the measurement in the ear.⁴

Measuring with a PPG in-ear sensor therefore offers a non-invasive and continuous method of measuring blood pressure. This also provides for the patient: in increased comfort: A non-invasive, everyday measurement of blood pressure is much more comfortable than conventional methods with cuffs, which, although automatic, contract every few minutes and can feel uncomfortable.

In recent years, cosinuss° has collected and comprehensively analyzed high-quality data in numerous studies – both internally and in collaboration with clinical partners. Based on this data, the machine learning models could be continuously trained and optimized. This has resulted in a valuable treasure trove of data with information from more than 500 people, over five million heartbeats and almost 60 days of recording. The studies included different scenarios and groups of people to ensure the most realistic and diverse data profile possible. Different measurement aspects were taken into account:

• Static blood pressure monitoring: Recording blood pressure values ​​in defined rest phases. The postures and measurement methods were based on the ISO 81060-2 standard.
• Dynamic blood pressure monitoring: Examination of blood pressure changes during physical exertion (e.g. running, cycling, everyday activities) and during temperature fluctuations – including measurements during surgical procedures.
• Reference measurements: Perform arterial measurements or, if this was not possible, cuff measurements as a basis for comparison.

Abb. 2: Comparison of systolic blood pressure measurements using c-med° alpha-Predictive model and a standard cuff measurement method over time. The solid line represents the c-med° alpha-Predictions while the dashed line shows the cuff measurements. The x-axis indicates time and the y-axis indicates systolic blood pressure in mmHg.

In recent years, cosinuss° has collected and comprehensively analysed high-quality data in numerous studies – both internally and in collaboration with clinical partners. The resulting database has made it possible to expand the neural network and continuously train and optimise the machine learning models in order to finalise a blood pressure algorithm that is already ready for approval. Next, cosinuss° will validate this algorithm in a clinical study and finalise the approval to be able to offer blood pressure as an additional certified vital parameter.