The world’s fastest antibiotic testing for neonatal units.
Halo AI tests a positive blood culture against the 8 antibiotics used in neonatal units and gives a susceptibility result in about 90 minutes. Standard testing takes two days.
Faster testing could help save up to 214,000 newborns a year.
Bring Halo AI to your lab Read the study
We have just raised £750k in seed funding to build the benchtop Halo Reader, and we’re hiring our first four people.
Amoxicillin still dividing
Gentamicin stopped
- Still dividing: the antibiotic isn’t working.
- Stalled and swelling: the antibiotic is working.
Run time from loading the cartridge, compared with standard testing. The bars are to scale.
It takes two days to find out whether a newborn’s antibiotics are working.
A baby born six weeks early develops a fever on day five, and overnight her blood culture flags positive. The team already has her on standard antibiotics. Whether those are the right ones won’t be known for up to two more days.
- 3 million newborns develop sepsis each year.
- Up to 570,000 of them die.
- An estimated 214,000 of those deaths are attributed to resistant bacteria. Faster testing could help prevent them.
Sources: Bulletin of the WHO, 2022; PLOS Medicine meta-analysis, 2021.
How it works
-
The instrument
The Halo Reader
A benchtop instrument that takes a single-use cartridge. The test itself already runs on a laboratory prototype, and we are now building the first Halo Readers.
-
Inside
Inside are phase-contrast optics and a camera, a heater that holds the cartridge at 37°C, a GPU that runs the AI on board, and the connection to your laboratory information system.
-
On the bench
The Halo Reader is designed to sit next to your blood culture incubators and run 24 hours a day.
-
The cartridge
The cartridge dilutes the sample, removes most of the blood cells and spreads the bacteria across about 500 microchannels. Eight sets of channels each hold one antibiotic. A ninth set has none and is the growth control.
-
Running a test
Step 1: Load
Pipette 50 µL of broth from the positive bottle into a cartridge and load it. Hands-on time is about 3 minutes.
-
Step 2: Watch
The Halo Reader keeps the cartridge at 37°C and photographs every channel once every 2 minutes.
-
Step 3: Read
The AI follows each cell from one image to the next. If the cells in a channel keep dividing, that antibiotic isn’t working. If they stall and swell, it is.
-
Step 4: Report
Halo AI runs alongside the mass spectrometry or molecular panel your lab already uses to identify the organism, which takes 35 to 65 minutes. It then applies the EUCAST breakpoints for that organism and sends the result to your laboratory information system.
Antibiotics on the panel: Benzylpenicillin, amoxicillin, flucloxacillin, gentamicin, cefotaxime, vancomycin, meropenem, amikacin.
How the AI reads a channel
Version 2.3.1 of the model was locked before our study in six NHS labs began, so every one of the 1,214 cultures in it was new to the software.
-
Find
A neural network boxes every cell in each image, including the chains and clusters that Gram-positive cocci form.
-
Follow
It links each cell to the same cell two minutes later, and to its two daughters when it divides.
-
Decide
A second model reads the tracks and classes the channel as still dividing or stopped. Where it can’t tell, it gives no call and recommends standard testing.
Built for the bacteria that infect newborns
Most rapid susceptibility systems test Gram-negative bacteria only. In newborns a lot of sepsis is Gram-positive, so we trained Halo AI on both.
Gram-positive, validated
- Group B streptococcus
- Staphylococcus aureus
- Coagulase-negative staphylococci
Gram-negative, validated
- E. coli
- Klebsiella
- Enterobacter
Tested on 1,214 blood cultures from newborns in 6 NHS labs.
-
96.8%
Agreement with reference broth microdilution.
-
0.7%
False-susceptible results. The accepted limit is 1.5%.
-
88min
Median time to result.
-
93.4%
Cultures reported within 2 hours.
Roadmap
-
Now
Proven technology
The cartridge, the imaging and the AI all work on a laboratory prototype, which we tested in 6 NHS labs. We have designed the benchtop Halo Reader but haven’t built one yet.
-
Next
Build the Halo Reader
We raised £750k, led by Kestrel Point Capital, to build the Halo Reader, hire our first four people and start taking the test into hospital labs.
-
Launch
Newborn sepsis
The first Halo Readers will go to NHS labs that serve neonatal units, where they will report on the 8 antibiotics used in newborns in about 90 minutes. Worldwide, an estimated 214,000 newborns die each year of sepsis attributed to resistant bacteria, and faster testing could help save them.
-
Then
Adult sepsis
Sepsis affects an estimated 49 million people of all ages each year, and 11 million of them die. An adult’s positive blood culture will run on the same cartridge and the same Halo Reader. Before we can offer it for adults we need to validate a longer list of species.
Source: The Lancet, 2020
-
Later
Urinary tract infections
There are an estimated 405 million urinary tract infections a year, and about 237,000 deaths. The Halo Reader, the imaging and the AI would stay as they are, and we would design a new cartridge that takes urine straight from the sample pot.
Source: Global Burden of Disease 2019
-
Long term
Other infections, starting with drug-resistant TB
Drug-resistant tuberculosis is first on our list, and about 390,000 people develop it each year. TB bacteria divide slowly and can only be handled in a biosafety-rated lab, so it will be a separate product.
We plan to place the Halo Reader free and charge £45 a test.
The plan is a reagent-rental agreement. We put a Halo Reader on your bench at no charge and you pay for the cartridges you use.
You can’t order one yet. If your lab serves a neonatal unit and processes more than 20 positive blood cultures a month, register and we’ll contact you when a pilot Halo Reader is ready.
Thanks, your lab is registered. We’ll reply within 2 working days.
So far Halo AI is one person, Ellie Rose. We’re hiring the next four.
Ellie Rose
Founder and CEO
Ellie Rose developed the method behind Halo AI in the Okoro Laboratory at the Sir William Dunn School of Pathology, University of Oxford, where she filmed 96 bacterial isolates against 8 antibiotics, 768 combinations in all. She is first author of the study in six NHS laboratories and reads biology at Corpus Christi College, Oxford.
Backed by
We raised a £750k seed round led by Kestrel Point Capital, a London investor in early-stage life sciences companies.
Frequently asked questions
What does Halo AI tell me?
It tells you whether the bacteria in a positive blood culture are susceptible or resistant to each of 8 antibiotics: benzylpenicillin, amoxicillin, flucloxacillin, gentamicin, cefotaxime, vancomycin, meropenem and amikacin. That covers the first-line combinations NICE recommends for neonatal infection (benzylpenicillin with gentamicin for early-onset, flucloxacillin with gentamicin for late-onset) and the drugs you would usually escalate to (NICE NG195; late-onset).
Does it identify the organism?
No, you carry on using whatever identification method you have now. Halo AI starts reading as soon as the cartridge is loaded and applies the right breakpoints when you enter the identification, so the final report comes within minutes of the ID.
How long does it take?
About 90 minutes from loading the cartridge. In the study the median was 88 minutes. It was quicker for E. coli (52 minutes) and slower for coagulase-negative staphylococci (102 minutes). Hands-on time is about 3 minutes, so a biomedical scientist on a night shift can run it.
Which organisms is it validated for?
The common neonatal pathogens: Group B streptococcus, Staphylococcus aureus, coagulase-negative staphylococci, E. coli, Klebsiella and Enterobacter. If the culture grows anything else, including fungi, Halo AI flags it for standard testing.
Can we have one now?
Not yet. So far the test has only run on a laboratory prototype, and we are using the seed round to build the benchtop Halo Reader. If you register your lab we will get in touch when pilot Halo Readers are ready.
What about mixed cultures?
Halo AI flags a mixed culture and recommends standard testing. It doesn’t come up often, because most positive blood cultures grow a single organism. In our study 21 of the 1,252 cultures tested were polymicrobial.
How accurate is it?
In our prospective study of 1,214 cultures at 6 NHS labs, Halo AI agreed with reference broth microdilution on 96.8% of results. It called a resistant organism susceptible in 8 of 1,116 resistant results, which is 0.7%. The accepted limits for a susceptibility test are at least 90% agreement and no more than 1.5% false-susceptible results, and our own target for agreement was 95%. The study ran on a laboratory prototype, so we will check the finished Halo Reader against these results before any pilot.
How many lives could it save?
Up to 214,000 a year. That is the estimated number of newborns who die each year of sepsis attributed to resistant bacteria (PLOS Medicine meta-analysis), and they are the babies a faster susceptibility test is for. It is an upper limit. Our study measured accuracy and speed and did not report results to clinicians, so it does not show an effect on treatment or outcomes. A randomised evaluation comes next.
Will it shorten antibiotic courses for babies who turn out not to be infected?
No. Halo AI only tests cultures that have flagged positive. A baby whose culture stays negative still follows NICE’s review at 36 or 48 hours to decide whether to stop. The benefit is for babies who are infected, who can be moved to the right drug sooner.
How will it fit into the lab?
The Halo Reader is a benchtop instrument designed to sit next to your blood culture incubators. It will send results to your laboratory information system and is meant to run 24 hours a day.
What will it cost?
We plan to charge £45 per test and to place the Halo Reader free of charge under a reagent-rental agreement.
Can we use it on adult cultures?
Not yet. We will validate it on adult cultures after the newborn launch.
Is it regulated?
Not yet. At the moment Halo AI is a research prototype and is not for clinical use. Before any result is used in patient care we will apply for UKCA marking as an in vitro diagnostic for susceptibility testing from positive blood cultures, and register the device with the MHRA.