Wearables for medical use need to acquire physiological signals with the highest quality and reliability. With the exponential deployment of digitalisation and big data, smart clothes are able to measure and process a large set of signals continuously, where the patient’s usual activity and environment will play a central role, since current medical practice lacks data that is measured in ‘the heart of the action’. Thus, data is not only useful for diagnostics or detecting pathologies, but may also help to determine the right dose of medication for a ‘personalised medicine’ approach, or to improve the medical knowledge and human body modelling.

Yet, smart clothes must be comfortable to wear, as well as cost-effective. It is well known that the monitoring of ECG (electrocardiogram) in ambulatory settings is routinely performed with a device called a Holter monitor, named after the biophysicist Norman Jefferis Holter, who patented his invention in 1965. Since then, Holter devices have seen their size reduce and autonomy increase, but their principle has not evolved.

Today, disposable adhesive-gel electrodes are electronically connected to a pocket recorder with shielded cables. Three to ten – or more – electrodes can be used, depending on the number of ECG channels, called leads, that must be recorded. The cables, which are often secured to the body with adhesive tape strips, result in a messy and not very comfortable set-up. Moreover, the relatively loose cables can unintentionally be pulled, causing signal artifact and disconnecting the electrode. It’s no wonder why engineers and product designers tried to improve the wearability of ambulatory ECG monitoring.

What is more surprising is that their commercial success was very limited, and the conservativeness of the medical industry is not the only place to lay blame; integrating an ECG set-up that is equivalent to a Holter monitor in a wearable, like a shirt, is indeed challenging.

A wearable Holter

One of the first devices to integrate a Holter in a cloth was, perhaps, the LifeShirt from VivoMetrics. However, it was not very ambitious, as the shirt was merely used to guide and secure the cables. Its adhesive-gel electrodes were applied on the patient’s body through slits in the shirt.

Replacing the adhesive-gel electrodes with dry contact ones is a huge improvement, in terms of comfort and use, especially when they are made from a conductive fabric. Unfortunately, signal quality is severely affected, making the system useless for medical applications beyond simple heart rhythm monitoring.

Some systems use clever methods to keep the textile electrodes humid, like the Cardionaute shirt from BioSerenity. Attempts have also been made to replace the shielded cables with textile connections. However, using a driven shield per cable must be kept to maintain sufficiently high immunity, with respect to electromagnetic disturbances, as well as good motion artifact rejection. Difficulty also arises by connecting with the recorder, as two contacts are needed per cable/electrode: one for the potential to measure and one for the driven shield. A recorder can easily have between seven and 20 connections with the shirt.

So far, no one has succeeded in making ECG measurements truly wearable, based on integrating this technology in a shirt. Existing solutions have to compromise signal quality, or they are not commercially cost-effective or sufficiently reliable. Their medical applications are therefore limited and are not able to fully replace existing Holter devices.

Cooperative sensors

With CSEM’s patented cooperativesensor technology, electrode sensors replace shielded cables and are connected in parallel to a bus. They can easily be integrated into simple conductive garments, lowering cost and creating comfortable yet powerful medical wearables. This is a radically different approach for measuring ECG that addresses the constraints of wearables.

Every sensor amplifies the biopotential directly at the electrode, producing high-input impedance that is suitable for dry electrodes. The signal is then transformed via modulation and sent to one master sensor (typically the one with the right-leg electrode) that records, processes and transmits all signals. Modulation may take the form of a digital signal if digitalisation is performed at the sensor itself.

One of the cooperative sensors’ unique features is their ability to be connected easily without losing signal quality. In their most advanced variants, the connection is limited to a single electrical line with no shield or insulation. The same line is used by the master sensor to set a reference of potential that, in turn, provides a path for current return to synchronise and communicate in both directions with all sensors.

Of course, a single line with no shield or insulation is easy to make with textile fabric. The other benefit of this approach is that every sensor is connected with the fabric via a single point. In contrast with the classical approach, the complexity of connection is independent of the number of sensors.

Cooperative sensors can easily be extended for other signals, like impedance. Multichannel-impedance modality, such as EIT (electrical impedance tomography), need classically double-shielded cables for best results. The signal quality obtained with cooperative sensors is not reduced, despite the fact that no shield and insulation is needed. As cooperative sensors include electronics at the measuring site, it is also relatively easy to acquire other kinds of signals, including thermal, optical, chemical, acoustic and kinematic types.

Furthermore, as the number of sensors can be large, without becoming more complex, maps can be made to pave the way for another kind of wearables that are able to take multidimensional signals. Functional images, for instance, are obtained via EIT and can be used to assess pulmonary edema.

The same electrodes can also be used in conjunction with ECG imaging (ECGi) to gather precise information on the heart’s electrical activity.

Finally, combining signals of a different nature provides richer information, enabling the computation of meta signals, such as SpO2 (pulse oxymetry), cuffless blood pressure or stroke volume.

Easy integration

Depending on different variants, the electronic part of cooperative sensors may require a substantial increase in technology compared with the classical approach.

However, once mastered, this technology is easily integrated in conventional integrated circuits. Combined with their ability to be integrated into textiles, they have the potential to provide medical-grade, powerful wearables at a low cost.

By reducing the sensors’ electronics to the size of dice and correctly encapsulating them, the smart clothes can be washed and disposed of with its electronics, making them simple to use. For this purpose, an ASIC has been developed that contains two main blocks: communication and sensing. The communication block can be configured as either master or slave. In the first case, a synchronisation data stream is sent out that consists of pulses that are generated by an internal clock source. For the latter, a dual-loop clock and data recovery (CDR) block extracts the sampling clock from incoming data pulses sent by the master.

In the digital TX part, data is semirandomised to create uniform spectrum that is applied prior to Manchester encoding. For every 1,000th bit, the start of a new frame is indicated by a violation in the Manchester code (‘code break’). Finally, the data is pulse modulated; for example, data ‘1’ is transformed into a positive pulse that is followed by a negative one. Data '0' is transformed into a negative pulse followed by a positive pulse. Detected pulses are demodulated, Manchester-decoded and descrambled in the digital Rx part. Code breaks are then used to synchronise the data frame.

Within the ASIC’s sensing part, the ECG is obtained from a measurement chain that consists of a preamplifier, passive low pass filter, instrumentation amplifier and an analogue-to-digital converter (ADC). The signal gain (26–40dB) and the hi/low pass-cutoff frequency (0.05–150Hz) of the preamplifier are set by external components.

The impedance measurement (IMP) incorporates two functions: stimuli and measurement. For stimuli, a currentinjection block delivers a modulated current in the 50kHz band. It consists of a modulator, filter and voltage-tocurrent converter, by means of an external resistor. The measurement chain consists of a preamplifier with a filter, demodulator, passive filter and an ADC. In this chain, in band signal (49–51kHz) is amplified and out of band signals are suppressed.

In order to guarantee that the ECG’s and impedance’s measurements will not be sensitive to motion artifacts or temporary changes in skin conditions, the impedance between body and garment electrode has to be kept as high as possible, posing a challenge to the front-end part of the sensing block. Another challenge is to guarantee that the direct current (DC) between electrodes will be close to zero to minimise electrolysis below the body’s electrode. For the dynamic range of measured signals, care must to be taken to limit noise feed-through from the communication band into the ECG/IMP band by demodulation.

Thanks to the developed ASIC, an excellent level of quality for the ECG and transthoracic impedance can be obtained with dry electrodes – at rest or in motion.

Cooperation is key

Increasingly complex demonstrators have been developed to show that the using cooperative sensors is effective. Verification from medical standards and clinical validations confirms that ECG and transthoracic electrical impedance can be measured with dry electrodes and simultaneously for many channels.

The validated sensors also included other sensing components – including an accelerometer, NTC thermistor, LED/ photodiode and an electret – that allow on-board algorithms to compute a large number of secondary signals, namely heart rate, core body and skin temperatures, SpO2, cuffless blood pressure, activity classes, energy expenditure, running speed and cough counting. The processing of signals is performed at the sensors to reduce the amount of data that is transferred.

Cooperative sensors are a new way to address multimodal measurements that are integrated into wearables. Thanks to their unique features, they reduce connection complexity without compromising signal quality. They can be deployed in large numbers and offer functional imaging capabilities.

In comparison with other wearables, this measurement approach has been reworked to address the constraints of wearables, rather than coping with the difficulties of integrating the classical Holter approach. Therefore, cooperative sensors can be manufactured with conventional technology (sensors and garments). Lower cost and more reliability can be achieved by taking advantage of the existing process. Today, the technology is made possible by integrating it into an ASIC.