Would be interesting how much resolution they need for the diagnosis to be reliable.
Because high resolution mass spectrometers cost millions of dollars, and "minutes" for a diagnosis can mean that one spectrometer can only run 3 samples per hour - or 72 per day.
And while a research university can afford a million dollar spectrometer (and the grad students that run it), even a small hospital will create 72 bacterial swaps per hour - while absolutely not having the money to get 10 spectrometers with the corresponding technicians.
And the incumbent/competitor - standard bacterial cultures - is cheap!
But couldn’t a hospital use it only for the critical cases? Where it is immediately important to know what infection someone may have, at imminent risk of death?
Who cares if they need to charge $30k per patient to use it if that fast knowledge saves the patient’s life. It doesn’t need to fully replace the existing methods, but could be a useful supplement when a patient is in critical condition
Soon resident here: not really super critical to know the exact pathogen. There are classes of antibiotics and depending on many factors we can have a good estimate of what will work so the first try usually helps. What's usually done (at least in france) is:
1. Take some blood to send to the labs to grow bacterias
2. Only then IV of antibiotics
3. Adjust at day 3 when you get the results.
And what if it looks super severe en urgent? Then 1. we do that too because it's often impossible to know what the pathogen was once you IVed all the antibiotics. And 2. we start by wide spectrum super strong high dose antibiotics.
Example: purpura fulminans: 1g of 3rd gen cephalosporins.
Ymmv in other countries, especially the USA because IIRC you have selected tons of emergent pathogens.
This. Literally last week, I had to go to the ER and they ran a PCR test to identify the infection. But they also started broad spectrum antibiotics immediately. But since it was a Thursday and the lab only works weekdays, I couldn’t get the results back until Tuesday. They adjusted the antibiotics to a more targeted one once the results were back.
Pretty usual stuff I'd say. And I'm willing to bet that the 2nd sets of antibiotic had a lower spectrum of action than the 1sr because we had some guess about the initial strain.
Also because you're still with us so it probably worked.
In any case I'm glad you're better and sorry you had to go through this.
Sure, but then they'd have a million(s) dollar machine (plus the people trained to operate it) waiting for these critical cases. It's a cost / benefit tradeoff, just like whether to use an MRI or CAT scan vs x-rays or an echo.
It doesn't have to be used for every patient, and it doesn't have to be kept around waiting for Doctor House's third act. There is nothing stopping you from a middle ground.
MRI machines cost up to 0.5 million, and take more than minutes for a scan. So this is in the upper realm of reasonable. At this point it is an engineering problem to get costs down.
> [...] have successfully demonstrated that a beam of light can not only be confined to a spot that is 50 times smaller than its own wavelength but also “in a first of its kind” the spot can be moved by minuscule amounts at the point where the light is confined.
Do you know how cheap culturing bacteria is ( I really have no idea )?
At 1 million/year ( worst case ), that’s roughly 3k/day so 40 dollars / analysis using your numbers . Assuming the machine lasts a bit longer ( say 5 years ) that’s 8 dollars/analysis, assuming 20min/sample. Make it a bit faster (5 minutes) and you’re down to two dollars which doesn’t look super expensive to me.
I would also wonder whether cultures are compatible with mass spec, and what the bottleneck is on measurement.
Culturing for optical microscopy seems like you need 1-10 DAYS before you get enough of a colony going to identify things, assuming you can successfully culture them at all. What if you plopped your culture down on the fermenter and gave it three hours (very cheap! Parallelizable!), then threw it in the mass spec? Would that shorten mass spec measurement time?
That’s assuming the machine works 24/7 matching marching perfectly the demand. I’m not a doctor but I guess it’s a seasonal "business" with more demand when cold winter starts, back to school September, epidemic episodes…
It could be very useful for Oncohematology patients. Cultures take a long time, relatively, for them, who often are severely immunosuppressed due to treatment.
If they are just looking for a few signature metabolites, you can easily do that on a single quadrupole for <$100k. Realistically, I would expect something more precise for medical purposes. Say a QExactive orbitrap -gives you phenomenal performance (<5ppm accuracy) and can be had for $500k.
The only mass specs I think of costing multiple millions of dollars are accelerator mass spectrometers (AMS) which are meant for isotopic analysis.
Mass spectromity is a bit expensive. My company can detect a common plant virus reliable with multi-spectral imaging only. Normal hires B/W cameras. 10 days before visible symptoms. SpexAI.com
Only vision scales. We did mass spectromity in Formula 1 because we had the money and it doesn't need to scale.
Such approaches are cool for plants and maybe animal healthcare, where the other reasonably priced alternative is nothing.
I don't know how I'd feel about reliable precision tools in human healthcare being substituted with B/W cameras and AI. It reminds me of a certain car company, where the human lives lost aspect wasn't so liked, even if they saved some money.
You forgot multi-spectral. That means single hires images for a lot of frequency ranges, which is better than a single color image, which merges together all frequencies at once.
In the end the accuracy of detections, false-positives, false-negatives wins. And this not comparable to this certain car company which refuses to use proper sensors. we also do infrared, 3d pointclouds to get the angle at each pixel, and thermal imaging.
another neighbor of us is doing real-time imaging for face detection, which is the industry leader world-wide. They can install a lot of cheap cams in football stadiums and give you a list of all 80.000 people, when they have access to the government passport/driver license photo database. they have.
but they cannot detect virus infections as we can do. nor bacterial infections. only if they do something illegal. who threw a bengal fire or started a fight.
You’re right, I didn’t notice the multi-spectral part. My point still stands, I wouldn’t want human diagnostic standards to slip but it is exciting technology. Thanks for sharing.
This. Mass Spectrometry should portably only be used to confirm/train AI vision models. Once a large enough data set has been created for a particular pathogen, incremental diagnostic cost would be effectively 0.
A reason for bw cam is sensitivity. No bayer filter means you capture all the photons. There are cameras that are sensitive enough to count photons as well. When I was in the lab we used EMCCD cameras (electron multiplier ccd) that we used to detect very low signals from cells (we wanted to be nice to them, if you hit them with UV light they don't like that).
There are all kinds of really great hacks to eke out an extra couple % at this level, if you consider you're counting photons. Back thinned, reverse mounting, high bandwidth multichannel readouts. Most of those trade offs are a thing of the past, but in the 2000s we were hacking the beans off those cameras to get them to do stunts for us.
It could be used for disease control and not necessarily personal health care. You could use something like the plant thing if it worked for people by photographing an entire stadium of people and finding infected people, airport entry lines, etc.
This is doing correctly what the car company did wrong. They start with a system that does not have a fatal incident upon failure. Meaning that even 1% or 10% failure rate is totally fine and possibly positive in terms of ROI.
On the other hand, even 0.01% is not acceptable when the system is responsible for human lives.
I also believe that cameras and visible spectrum are scalable and sufficient, and there is evidence it is so. The most reliable system driving cars is the homo sapiens and apart from a bit of proprioceptive, vehicles are operated based on vision sensors(eyes).
To conclude, I find it a clever move to start refining the technology in a field where ROI is in dollar terms and not in lives saved.
>On the other hand, even 0.01% is not acceptable when the system is responsible for human lives.
I've got bad news for you about a whole lot of medical tests and interventions if 0.01% is unacceptable for you. I get your point that the stakes might be higher with medical technology than in some other fields, but you're still setting the bar too high.
The first pathogen I saw mentioned by name while skimming through the linked article was H. Pylori, where the false-positive and false-negative rates for various currently used tests are several dozen times worse than that at about 0.30-0.90% [0]. Even just being hospitalized in and of itself has failure rates. Just eyeballing some of the numbers at [1] and elsewhere, I suspect that at least 0.01% of hospitalizations result in a death attributable to a hospital acquired infection.
You can identify quite a few pathogens with Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy. I think the machines for that are smaller and cheaper than mass spec machines.
"In recent years, matrix-assisted laser desorption/ionisation-mass spectrometry (MALDI-MS) has been rapidly adopted in clinical practice to identify bacteria based on unique protein profiles6,7. However, while the identification process itself is faster, MALDI-MS-based analysis suffers from the same time-consuming isolation step as traditional approaches. Time, however, is a decisive factor in the successful treatment of several infection scenarios such as sepsis4,8. Consequently, the ideal scenario of microbial diagnosis is to identify bacteria directly from the clinical sample. Using MALDI, so far only a small number of applications could be implemented due to the high protein background in most clinical samples9."
While mass spectrometry is an old(-ish) method, only recently has it become good and fast enough for clinical use. They also needed to make their own database of bacterial byproduct signatures to look for, as that did not exist yet.
As to "why just now": not everything has been invented yet, even things that might seem relatively straightforward.
Article mentions existing tools e.g., MasSpec Pen, which detect signals of interest in samples with minimal processing (i.e., avoiding the hard part of processing/cleaning bio samples). But those were specific cancer signals that stood out from the general protein noise. (See also the cancer knife, that burns to cut and analyzes the smoke to see if you have enough margin).
Problem with bacteria is (a) wide diversity and (b) previous work was mostly on lipids, not small molecules. For small molecule metabolites of bacteria, previous article reported a medium-size proof of concept (hundreds of target), and this is doing more (thousands), where they identify a few hundred unique patterns and then demonstrate their detection.
At 90%, it's likely this would be used for pre-screening to identify possible candidates, though it might reinforce initial clinical suspicion and preliminary antibiotics.
Incredibly capable mass spectrometers can be had for half a million. Plenty of undergrad labs have access to one. A student quality single quad is <$100k.
Very possible they're doing the research now, because one of the mass spectrometer companies has gotten ISO 13485 certified and can now offer the spectrometer for diagnostics in humans as a medical device manufacturer.
"Our innovative approach is not to look directly for the pathogenic bacteria, but only for their metabolic products. This allows us to detect them indirectly, but much more quickly."
Would be interesting how much resolution they need for the diagnosis to be reliable.
Because high resolution mass spectrometers cost millions of dollars, and "minutes" for a diagnosis can mean that one spectrometer can only run 3 samples per hour - or 72 per day.
And while a research university can afford a million dollar spectrometer (and the grad students that run it), even a small hospital will create 72 bacterial swaps per hour - while absolutely not having the money to get 10 spectrometers with the corresponding technicians.
And the incumbent/competitor - standard bacterial cultures - is cheap!
But couldn’t a hospital use it only for the critical cases? Where it is immediately important to know what infection someone may have, at imminent risk of death?
Who cares if they need to charge $30k per patient to use it if that fast knowledge saves the patient’s life. It doesn’t need to fully replace the existing methods, but could be a useful supplement when a patient is in critical condition
Soon resident here: not really super critical to know the exact pathogen. There are classes of antibiotics and depending on many factors we can have a good estimate of what will work so the first try usually helps. What's usually done (at least in france) is:
1. Take some blood to send to the labs to grow bacterias
2. Only then IV of antibiotics
3. Adjust at day 3 when you get the results.
And what if it looks super severe en urgent? Then 1. we do that too because it's often impossible to know what the pathogen was once you IVed all the antibiotics. And 2. we start by wide spectrum super strong high dose antibiotics.
Example: purpura fulminans: 1g of 3rd gen cephalosporins.
Ymmv in other countries, especially the USA because IIRC you have selected tons of emergent pathogens.
This. Literally last week, I had to go to the ER and they ran a PCR test to identify the infection. But they also started broad spectrum antibiotics immediately. But since it was a Thursday and the lab only works weekdays, I couldn’t get the results back until Tuesday. They adjusted the antibiotics to a more targeted one once the results were back.
Pretty usual stuff I'd say. And I'm willing to bet that the 2nd sets of antibiotic had a lower spectrum of action than the 1sr because we had some guess about the initial strain.
Also because you're still with us so it probably worked.
In any case I'm glad you're better and sorry you had to go through this.
Sure, but then they'd have a million(s) dollar machine (plus the people trained to operate it) waiting for these critical cases. It's a cost / benefit tradeoff, just like whether to use an MRI or CAT scan vs x-rays or an echo.
Plus high-res mass specs can't just be turned on, they need to be kept on standby which isn't cheap when you cost it out over a year
This is a false dichotomy.
It doesn't have to be used for every patient, and it doesn't have to be kept around waiting for Doctor House's third act. There is nothing stopping you from a middle ground.
In the US, this is already handled by prior authorization.
/s
MRI machines cost up to 0.5 million, and take more than minutes for a scan. So this is in the upper realm of reasonable. At this point it is an engineering problem to get costs down.
A good MRI machine is way more than $500k!
There are 0.05 Tesla MRI machines that almost work with a normal 15A 110V outlet now FWIU; https://news.ycombinator.com/item?id=40965068 :
> "Whole-body magnetic resonance imaging at 0.05 Tesla" [1800W] https://www.science.org/doi/10.1126/science.adm7168 .. https://news.ycombinator.com/item?id=40335170
Other emerging developments in __ spectroscopy:
/?hnlog Spectro:
NIRS;
> Are there implied molecular structures that can be inferred from low-cost {NIRS, Light field, [...]} sensor data?
NIRS would be low cost, but the wavelength compared to the sample size.
From https://news.ycombinator.com/item?id=38528844 :
> "Reversible optical data storage below the diffraction limit (2023)" [at cryogenic temperatures] https://news.ycombinator.com/item?id=38528844 :
> [...] have successfully demonstrated that a beam of light can not only be confined to a spot that is 50 times smaller than its own wavelength but also “in a first of its kind” the spot can be moved by minuscule amounts at the point where the light is confined.
"Eye-safe laser technology to diagnose traumatic brain injury in minutes" https://news.ycombinator.com/item?id=38510092 :
> "Window into the mind: Advanced handheld spectroscopic eye-safe technology for point-of-care neurodiagnostic" (2023) https://www.science.org/doi/10.1126/sciadv.adg5431
> multiplex resonance Raman spectroscopy
Holotomographic imaging is yet another imaging method that could be less costly than MRI; https://news.ycombinator.com/item?id=40819864
"Quantum microscopy study makes electrons visible in slow motion" https://news.ycombinator.com/item?id=40981054 :
> "Terahertz spectroscopy of collective charge density wave dynamics at the atomic scale" (2024) https://www.nature.com/articles/s41567-024-02552-7
Do you know how cheap culturing bacteria is ( I really have no idea )?
At 1 million/year ( worst case ), that’s roughly 3k/day so 40 dollars / analysis using your numbers . Assuming the machine lasts a bit longer ( say 5 years ) that’s 8 dollars/analysis, assuming 20min/sample. Make it a bit faster (5 minutes) and you’re down to two dollars which doesn’t look super expensive to me.
I would also wonder whether cultures are compatible with mass spec, and what the bottleneck is on measurement.
Culturing for optical microscopy seems like you need 1-10 DAYS before you get enough of a colony going to identify things, assuming you can successfully culture them at all. What if you plopped your culture down on the fermenter and gave it three hours (very cheap! Parallelizable!), then threw it in the mass spec? Would that shorten mass spec measurement time?
That’s assuming the machine works 24/7 matching marching perfectly the demand. I’m not a doctor but I guess it’s a seasonal "business" with more demand when cold winter starts, back to school September, epidemic episodes…
Aren't there are also operating cost ? What you said is just depreciating cost of the machine.
I wonder if it needs the high resolution models for this technique? The models we used in undergrad were far cheaper (probably $250k and up).
It could be very useful for Oncohematology patients. Cultures take a long time, relatively, for them, who often are severely immunosuppressed due to treatment.
If they are just looking for a few signature metabolites, you can easily do that on a single quadrupole for <$100k. Realistically, I would expect something more precise for medical purposes. Say a QExactive orbitrap -gives you phenomenal performance (<5ppm accuracy) and can be had for $500k.
The only mass specs I think of costing multiple millions of dollars are accelerator mass spectrometers (AMS) which are meant for isotopic analysis.
Mass spectromity is a bit expensive. My company can detect a common plant virus reliable with multi-spectral imaging only. Normal hires B/W cameras. 10 days before visible symptoms. SpexAI.com
Only vision scales. We did mass spectromity in Formula 1 because we had the money and it doesn't need to scale.
Such approaches are cool for plants and maybe animal healthcare, where the other reasonably priced alternative is nothing.
I don't know how I'd feel about reliable precision tools in human healthcare being substituted with B/W cameras and AI. It reminds me of a certain car company, where the human lives lost aspect wasn't so liked, even if they saved some money.
> B/W cameras and AI.
You forgot multi-spectral. That means single hires images for a lot of frequency ranges, which is better than a single color image, which merges together all frequencies at once.
In the end the accuracy of detections, false-positives, false-negatives wins. And this not comparable to this certain car company which refuses to use proper sensors. we also do infrared, 3d pointclouds to get the angle at each pixel, and thermal imaging.
another neighbor of us is doing real-time imaging for face detection, which is the industry leader world-wide. They can install a lot of cheap cams in football stadiums and give you a list of all 80.000 people, when they have access to the government passport/driver license photo database. they have. but they cannot detect virus infections as we can do. nor bacterial infections. only if they do something illegal. who threw a bengal fire or started a fight.
You’re right, I didn’t notice the multi-spectral part. My point still stands, I wouldn’t want human diagnostic standards to slip but it is exciting technology. Thanks for sharing.
This. Mass Spectrometry should portably only be used to confirm/train AI vision models. Once a large enough data set has been created for a particular pathogen, incremental diagnostic cost would be effectively 0.
A reason for bw cam is sensitivity. No bayer filter means you capture all the photons. There are cameras that are sensitive enough to count photons as well. When I was in the lab we used EMCCD cameras (electron multiplier ccd) that we used to detect very low signals from cells (we wanted to be nice to them, if you hit them with UV light they don't like that).
There are all kinds of really great hacks to eke out an extra couple % at this level, if you consider you're counting photons. Back thinned, reverse mounting, high bandwidth multichannel readouts. Most of those trade offs are a thing of the past, but in the 2000s we were hacking the beans off those cameras to get them to do stunts for us.
Things are fast good AND cheap now though!
It could be used for disease control and not necessarily personal health care. You could use something like the plant thing if it worked for people by photographing an entire stadium of people and finding infected people, airport entry lines, etc.
This is doing correctly what the car company did wrong. They start with a system that does not have a fatal incident upon failure. Meaning that even 1% or 10% failure rate is totally fine and possibly positive in terms of ROI.
On the other hand, even 0.01% is not acceptable when the system is responsible for human lives.
I also believe that cameras and visible spectrum are scalable and sufficient, and there is evidence it is so. The most reliable system driving cars is the homo sapiens and apart from a bit of proprioceptive, vehicles are operated based on vision sensors(eyes).
To conclude, I find it a clever move to start refining the technology in a field where ROI is in dollar terms and not in lives saved.
To the found: Good luck, i am jealous.
>On the other hand, even 0.01% is not acceptable when the system is responsible for human lives.
I've got bad news for you about a whole lot of medical tests and interventions if 0.01% is unacceptable for you. I get your point that the stakes might be higher with medical technology than in some other fields, but you're still setting the bar too high.
The first pathogen I saw mentioned by name while skimming through the linked article was H. Pylori, where the false-positive and false-negative rates for various currently used tests are several dozen times worse than that at about 0.30-0.90% [0]. Even just being hospitalized in and of itself has failure rates. Just eyeballing some of the numbers at [1] and elsewhere, I suspect that at least 0.01% of hospitalizations result in a death attributable to a hospital acquired infection.
[0] https://www.aafp.org/pubs/afp/issues/2019/0701/p16.html [1] https://www.cdc.gov/healthcare-associated-infections/php/dat...
Sounds like this is exploiting a correlation between the virus and how the plant reacts to it. That's very specific.
The nice thing about mass-spec is that you can use it as a dragnet detection method.
>Normal hires B/W cameras. 10 days before visible symptoms.
*visible to the human eye
You can identify quite a few pathogens with Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy. I think the machines for that are smaller and cheaper than mass spec machines.
see https://pmc.ncbi.nlm.nih.gov/articles/PMC5631018/ , and this company that is commercialising it https://microbira.com/
This is cool, but I still see sequencing and metagenomics as being the gold standard going forward.
Metagenomics can do hard diagnoses. We achieved the first truly confirmed diagnosis of neurohistoplasmosis in our region using that technique.
Yes specifically mcfDNA which also has the advantage of only needing a blood draw
Super interesting! But mass spectrometry is an old method, why just now?
Checking out the actual article (https://www.nature.com/articles/s41467-024-55457-7) the authors write that:
"In recent years, matrix-assisted laser desorption/ionisation-mass spectrometry (MALDI-MS) has been rapidly adopted in clinical practice to identify bacteria based on unique protein profiles6,7. However, while the identification process itself is faster, MALDI-MS-based analysis suffers from the same time-consuming isolation step as traditional approaches. Time, however, is a decisive factor in the successful treatment of several infection scenarios such as sepsis4,8. Consequently, the ideal scenario of microbial diagnosis is to identify bacteria directly from the clinical sample. Using MALDI, so far only a small number of applications could be implemented due to the high protein background in most clinical samples9."
While mass spectrometry is an old(-ish) method, only recently has it become good and fast enough for clinical use. They also needed to make their own database of bacterial byproduct signatures to look for, as that did not exist yet.
As to "why just now": not everything has been invented yet, even things that might seem relatively straightforward.
glossing...
Article mentions existing tools e.g., MasSpec Pen, which detect signals of interest in samples with minimal processing (i.e., avoiding the hard part of processing/cleaning bio samples). But those were specific cancer signals that stood out from the general protein noise. (See also the cancer knife, that burns to cut and analyzes the smoke to see if you have enough margin).
Problem with bacteria is (a) wide diversity and (b) previous work was mostly on lipids, not small molecules. For small molecule metabolites of bacteria, previous article reported a medium-size proof of concept (hundreds of target), and this is doing more (thousands), where they identify a few hundred unique patterns and then demonstrate their detection.
At 90%, it's likely this would be used for pre-screening to identify possible candidates, though it might reinforce initial clinical suspicion and preliminary antibiotics.
Old does not really mean widely available. These machines can run into the hundreds of millions of dollars.
Having one spare for research is not the norm, any more than having one is.
Incredibly capable mass spectrometers can be had for half a million. Plenty of undergrad labs have access to one. A student quality single quad is <$100k.
Sure, but those aren't high resolution - like this technique requires.
A QE is an absolutely fantastic instrument for half a million and more capable than the baseline Exactive used here.
Often, stuff like that is a compliance thing.
Very possible they're doing the research now, because one of the mass spectrometer companies has gotten ISO 13485 certified and can now offer the spectrometer for diagnostics in humans as a medical device manufacturer.
"Our innovative approach is not to look directly for the pathogenic bacteria, but only for their metabolic products. This allows us to detect them indirectly, but much more quickly."
I wonder how much resolution they need for diagnosis...