Enhancing Bluetooth Contact Tracing for COVID-19 Containement with Humidty/Pressure sensors

Update April 2021

The project has now successfully completed and the results are entering commercial marketing. We are pleased to have worked together with the UK government funding agency Innovate UK to deliver this novel technology.

Dedicated web-site launches!

We have received UK Government funding to develop this technology further, and all new developments will be posted on dedicated website: www.atmofencing.com.

Executive summary

Radio-based contact tracing, e.g., using Bluetooth, for suppressing COVID-19 suffers from inaccuracies as travel of radio waves is not necessarily representative of infection risks. We are developing technology to use wearable humidity sensors to augment the radio measurements with “atmo-fencing”, i.e., identifying how well the air surrounding contacts is mixed and therefore improve accuracy prediction of infection between contacts. More accurate contact tracing will increase public confidence and accelerate return to more normal social and economic activity.

Introduction

Current strategy for dealing with the COVID-19 pandemics is to physically interrupt the human chain of infection and, for the seriously sick, supportive hospital treatment.

We do not currently have sufficiently reliable ways of prospectively identifying either the infectious nor the susceptible population; prospectively identifying infectious individuals is difficult because physical symptoms (such as cough, fever) are not a reliable indicator and because mass screening is unavailable and potentially not reliable enough. Identification of immune individuals by and large1 will need to await a development of a sensitive and reliable anti-body test that can be used at a large scale.

As a consequence of the above, in much (perhaps most) of the world the strategy to interrupt the chain of infection is to assume all individuals are infectious and susceptible and introduce various enforced physical distancing measures to the whole population2. With infection rates high and serious cases overwhelming the hospital systems no other strategy is currently viable. However the current phase can be expected to bring the infection rates under control after which there will potentially be an opportunity for technology-based mechanisms (amongst others) for interrupting the transmission chains.

Bluetooth contact tracing and shortcomings

Digital contact tracing using Bluetooth and other technologies has been developed and used early on in the pandemic in Asia, and is now being proposed more widely, e.g., by Google and Apple. This could be a promising approach and while there are likely to be substantial privacy concerns it is likely that the privacy intrusion can be minimised sufficiently to make the approach acceptable for the duration of the pandemic (if not a permanent strategy for containing cold and flu viruses).

The premise of this technology is to keep a record of potential chains of transmission so that if an individual is diagnosed with COVID-19 during the course of their infection (either through symptoms or screening) the whole potential chain is informed of this and, based on this information, suitably isolated to prevent further transmission.

It is recognised that technology based on solely transmission/reception of cm-wave radio signals will inevitably have some shortcomings:

  1. Range of radio waves depends on various factors meaning that true geometric distance to contact is not reliably estimated

  2. Even if geometric distance were accurate, this is not a perfect indicator of likely infection, because:

    i. The receivers could separated by a radio-transparent barrier (thin wall, glass window a vehicle) which the virus can not cross

    ii. There degree to which the infectious aerosol of the virus is retained near the infectious person is not estimated

    iii. It is likely that infection rates depend on physical parameters of the air, i.e., temperature and humidity.

Enhancing Bluetooth tracing with humidity/pressure sensors

The shortcomings of pure radio-based contact tracing can be mitigated by augmenting the radio data with other information. Here we propose that information from air humidity and pressure sensors could be particularly useful.

“Atmo-fencing”

If two individuals (together with their humidity/pressure sensors) are situated so that the air they breathe is well-mixed then the air relative humidity and pressure are likely to be measured to be very close. However, if the air is poorly mixed (e.g. one is in a stationary vehicle while other is outside) then the physical parameters of the air are likely to diverge significantly. The humidity/pressure data can therefore be used as an atmospheric equivalent of “geo-fencing”, i.e., “atmo-fencing”.

Since droplet/airborne infection is thought to be the major COVID-19 infection route this should provide key information on infection risk.

Aerosol linger time

One of the likely routes of COVID-19 infection is through an aerosol of the virus particles. The concentration of the virus will depend on the ventilation of an enclosed space. In enclosed spaces with significant number of people present, relative humidity is an indication of ventilation rates.

The inference of the ventilation is substantially more complex then simple determining if the air is well mixed between two points, nevertheless a useful estimator should be possible. If such an estimator is developed then this should be a useful input into calculating the infection likelihood.

Physical influence on transmission

It is possible that air relative humidity has a direct influence on virus transmissible by changing the stable size of very fine droplets in which the virus is suspended in air. Hence relative humidity during contact time, on its own, may provide a useful input in calculating the likelihood of transmission.

Humidity sensors

Air pressure (i.e., barometric) sensors are present on some mobile phones. Air humidity sensors now seem very rare but some mobile phone models used to have them, e.g., Samsung Galaxy S4. Even if present, a mobile phone air humidity sensor may not give useful information, e.g., because mobile phones are often placed in cases, pockets or purses. It is likely a humidity sensor worn on outer layer of clothing would be needed for useful information.

Such wearable sensor technology is now highly advanced. For example, the Bosch BME280 provides highly accurate relative humidity and pressure sensing (equivalent to 1 meter in elevation!) at low power consumption. This sensor is already integrated in commercially available wearable modules which allows for investigation on suitability for COVID-19 contact tracing to start immediately.

What needs to be done

In reverse chronological order!

End goal: controlled study in the community

The end-goal is a study in the community to identify if augmenting radio-only with humidity and pressure data improves the retrospective prediction of COVID-19 infection. Deployment in high-risk environments (e.g., among community health workers) may accelerate collection of useful information.

Wearing a humidity sensor on voluntary basis is likely selective with respect to behaviour relevant to infection risk, and may modify future behaviour. A suitable protocol to control for these effects may be needed (e.g., informing subjects that the sensor may be a placebo that can not collect data).

Intermediate goal 2: Maintaining privacy protection with augmented data

A number of protocols have already been devised which maintain reasonable privacy while allowing contact tracing. These will need to be revised, or new protocols devised, which are able to maintain privacy while also using the augmented data from pressure/temperature asensors.

Intermediate goal 1: Study of proposed routes of effect

The intermediate goal is to study if practical humidity/pressure sensors can be used to infer the relevant factors which may influence transmission. In particular:

  1. Is it possible to accurately gauge how well the air is mixed between two individuals in practical settings?

  2. Is it possible to infer with reasonable accuracy the ventilation of a enclosed space?

  3. Literature study of influence of relative air humidity on COVID-19 infection rate

First steps

First steps are simple enables:

  1. Acquisition and evaluation of a number of commercially available ready-to-wear modules to select model(s) most suitable for further study

  2. Development of data acquisition and analysis software, initially off-line on a computer and subsequently on a mobile phone platform (or platforms).

How you can help?

We are just getting going on this and it is currently an unfunded concept. Areas where help would be appreciated, in currently anticipated order of usefulness:

  1. Comments/Feedback
  2. Cash grant funding
  3. Offers to collaborate
  4. Contribution to the software development effort
  5. Donation of ready-to-wear accurate pressure/humidity sensors

For time being please send any offers/enquiries to webs@bnikolic.co.uk

Version history

Date Version Comment
11/4/2020 1.0 Initial public version
21/4/2020 1.1 Add executive summary
24/5/2020 1.2 Add link to new website

Footnotes

  1. We probably can already assume that persons who displayed symptoms, were positively tested for COVID-19 and then recovered will be immune from re-infection for at least some time. However, the fraction of COVID-19 cases confirmed in this way is too low to usefully influence public policy. 

  2. Different measures are sometimes applied different age groups