Different Applications of our Device

A Diagnostic Tool

The Global Antimicrobial Resistance Issue

When choosing an issue to tackle with our diagnostic hardware, we decided to focus on the growing issue of antimicrobial resistance (AMR). AMR infections are thought to claim as many as 700,000 lives per year, a figure that is expected to rise to 10 million by 2050 if current trends continue (Figure 1) (1). The World Health Organisation (WHO) has therefore described AMR as one of the major concerns of modern healthcare (2), and it is critical that we take action now.

The use of antibiotics is the leading cause of antimicrobial resistance around the world, and in 50% of cases, antibiotic treatment is either not needed or not optimally effective for the infection (3). Doses of antibiotic that are not lethal to bacteria leads to the development of antibiotic resistance by supporting genetic change and acting as a selection pressure (4) (Figure 2). Therefore, one way clinicians can combat AMR is to conduct antimicrobial susceptibility tests (ASTs). These identify any drug resistance in pathogens and ensure effective drugs are chosen for treatments (5).

Phase I: Learning from Microbiology labs

To gain a better understanding of what ASTs are most commonly used in the clinic, we spoke to microbiologist consultants in Poland and Cyprus. They informed us of the Vitek 2 System, that is commonly used for ASTs.

Vitek 2 is a highly automated system, in which a bacterial suspension grown from a patient sample, is distributed into compact cards containing antibiotics (figure 3). A sensitive optical density system then measures changes in bacterial growth in the presence of the antibiotics in the cards. If there is no bacterial growth, then the organism is sensitive to the drug and it would therefore be an appropriate treatment option.

However, the Vitek 2 system is hugely expensive, costing close to $100,000 (7). The clinicians in Poland and Cyprus expressed that this cost was the major drawback of Vitek 2, and means that often smaller hospitals and clinics are unable to afford this state of the art machinery, and therefore miss out on the ASTs that could be crucial to prevent the spread of AMR.

Phase I: Integration of feedback

Following our research on the Vitek 2 system, we noticed that our device could have many of the same features. It detects bacterial growth as a measure of optical density, and so may therefore detect changes in growth at rates similar to Vitek 2. In addition, we could emulate the Vitek 2 cards by putting antibiotics in the well plate, and seeing how various antibiotics affect growth of a bacterial inoculum.

Firstly, we made sure we could detect differences in growth of antibiotic resistant and sensitive strains in the presence of the appropriate antibiotic, as Vitek 2 can. We engineered antibiotic resistant strains by transforming cells with biobricks encoding chloramphenicol, kanamycin, carbenicillin and ampicillin resistance. Then used a laboratory automated plate reader to compare their growth with that of of antibiotic sensitive strains, in the presence of antibiotics. We found that with the laboratory plate reader we could see a difference within just a few hours at bacterial concentrations as low as 10^3 CFU/ml (equivalent to 1x10-3 OD) (figure 4). This showed us that it would be possible to differentiate between antibiotic sensitive and insensitive strains, in a time frame that could be clinically relevant.

The next step will be to test our device for its ability to distinguish between an antibiotic sensitive and resistant strain, and ensure our results are comparable to the laboratory plate reader.

Phase II: Taking our device to healthcare professionals

We then took our Antimicrobial Susceptibility Test prototype to professionals working in clinics and diagnostic settings, the markets we imagine our device entering. We interviewed them about our prototype device, and asked for constructive feedback. We have attempted to incorporate as much of their feedback as possible into our current device design:

The Royal Hallamshire Hospital in Sheffield, England

We spoke to senior nurses on the wards of a major hospital in Sheffield. They were enthusiastic about the cost of our device, as it is significantly cheaper than the single use tests they currently use for infections such as MRSA screening. However, they felt that the ~8 hour test time of our device would slow down patient flow on wards. Instead, they suggested we contact diagnostic labs directly, whom the hospital uses to diagnose severe patients staying overnight on the wards.

Liverpool Diagnostics in Liverpool, England

We then turned to a major diagnostic lab in the North of the UK. They highlighted to us the importance of identifying the pathogenic organism, as this will be important in patient care, and also in choosing which antibiotics to test against the organism. For bacterial identification, we investigated several mechanisms such as microcalorimetric methods. This idea comes from research by Chang-Li et al. 1988, who showed that growing bacteria leave a very characteristic thermal finger print. In their study Chang-Li et al. were able to distinguish among Escherichia coli, Staphylococcus aureus, Salmonella typhosa, Salmonella chaleraeuis and Shigella flexneri using micro temperature sensors. Given eenough time, we would like to try and employ similar technology in our device. However this technology is largely eexperimental and we may therefore need to explore other options.

They also suggested we create a way to make our device more tailored to certain types of pathogen. To do this, we would mimic the Vitek 2 cards, and design well plates, pre-filled with antibiotic appropriate to a certain group of antibiotics. This will increase the autonomy and ease of use of our device, and as 96 well plates are already a standard consumable, they should be much cheaper to produce than the custom Vitek 2 cards.

PreventX Integrated Diagnostics, Sheffield, England

PreventX is a major diagnostics lab based in Sheffield. They told us it was important that our device had an integrated software that can analyse the sample and produce a definitive, easy to read outcome.
To do this, we would implement machine learning software, and use a large amount of bacterial growth data to ‘teach’ our software difference between ‘growth’ and ‘no growth’, i.e. sensitivity or resistance to the drug tested. It would then provide the user with an easily interpreted result. We have designed what this software might look like in figure 5.

PreventX were also concerned with patient confidentiality, as our device would be sending patient data over the cloud. To combat this issue, we spoke to various members of research ethics staff at the University of Sheffield, and decided that the best course of action would be to design our own encryption software.

The software uses a discrete mathematical model (based on cellular automata) similar to conway's game of life to generate a string of pseudo random numbers. It uses a JPEG image as an input, then performs image processing to convert it to a binary array and then performs the encryption algorithms on this data.This is then applied to data like a shift cypher and sent to the server. To decrypt the data the shift cypher is removed and the original data is obtained.

They also advised on best practices such as assigning patients anonymous identifier and then not sending information linking identifiers to their actual name through the same medium.