Pakistani Students Develop a Kit to Identify COVID Patients in 20 Seconds

[Muhammad Omar]

Pakistani Students Develop a Kit to Identify COVID Patients in 20 Seconds

Two Pakistani students have developed a device that can diagnose coronavirus through AI within 20 seconds.

Mohammad Aleem and Rahul Raj, students at Gulam Ishaq Khan Institute (GIKI), are the masterminds behind this invention.

The students say that the shortage of testing kits and long waiting time after the test motivated them to turn to technology amid growing coronavirus cases across the world.

A timely diagnosis would allow doctors to initiate the isolation and treatment process of a COVID-19 patient instantly, reducing the risk of cross-infection to a great extent.

How the Device Works
The AI-powered device developed by GIKI students uses computed tomography (CT) scan of lungs to detect COVID-19.

The device can perform analysis on CT scans and tell whether a patient has been infected with coronavirus in just 20 seconds with 92% accuracy.

Besides speed and accuracy, the device can also identify the exact location, impact, and severity of the damage inflicted on the lungs due to coronavirus.

Requesting Funds From The Government
The AI-powered device depends on CT scans in order to diagnose the disease. A CT scan costs somewhere between Rs. 5,000 to 10,000 in Pakistan. The cost of a CT scan is the biggest impediment to the mass production of the AI-powered device.

Moreover, the more data the AI-based machine processes, the more accurate it will become with time.

The students have requested the government to provide them adequate resources so they can move towards the mass-production of the device

 

[masterchief_mirza]

Excellent. The CT burn out method is how China successfully rapidly triaged thousands of suspected cases without even waiting for lab results. Coupling AI to this is a superb breakthrough and if validated, could well permit Pakistan to adopt an identical strategy for isolating and treating covid patients separately and more aggressively to those with other ailments.

The problem may be in capacity. I understand Karachi now has a purpose built hospital for covid patients. We need two or three such facilities in every metropolis, and along with this rapid testing method AND sufficient ventilators AND personnel to man them AND PPE to eliminate risk of Infection of those personnel, that combination of elements will see us through inshAllah.

Hope IK is noticing and coordinating these separately running projects.

Leadership is needed to bring this fragmented response together into one unified push.

 

[khail007]

Appreciation for concern and hard-work of both for humanity.
Device's diagnostic ability depends upon the CT-Scan results. So the mass production is conditional to CT-Scan facilities available at hospitals across the country.

 

[DocEinstein]

Very well done using your brains to try and come up with new and novel solutions.
However I am skeptical without knowing the specifics of the device and what is it looking for on CT results. 92% accuracy ?????
How is the sensitivity and specificity calculated without any test data collected on real cases?
CT scan testing is not a benign/ easy modality. It is not going to be readily available plus it does exposes you to a lot of radiation.
If the device is just looking for an ARDS like picture and calling it Corona virus related presentation then surely in today's pandemic overwhelming cases with those CT scan features would be secondary to corona virus, however in a patient with ARDS secondary to any other Viral or bacterial etiology the false positive rate will be 100%(in the absence of blood or nasopharyngeal swab testing for actual presence of virus) and in the interim may lead to wrong diagnosis and treatment.
Same is the case with false negative results. No pulmonary involvement with milder cases would have a high false Negative rate on such kind of testing with patient returning to general population and still spreading infection through contact.

 

[masterchief_mirza]

Appreciation for concern and hard-work of both for humanity.
Device's diagnostic ability depends upon the CT-Scan results. So the mass production is conditional to CT-Scan facilities available at hospitals across the country.

Big cities have lots of CT scanners in major hospitals. Repurpose them for rapid throughput. I am only saying this as this is what worked in China. It may not be doable in Pakistan but would be worth trying.

Countries that wait for the tsunami to hit and are reactive in their approach are being hit hardest. Being proactive and literally hunting for the diagnosis among anyone symptomatic - certainly in Karachi it is doable - is the best way forwards. This way we can isolate the patients immediately and start treating them, while keeping them from infecting others. Surely the Wuhan model must be applied to Karachi post haste.

S

Very well done using your brains to try and come up with new and novel solutions.
However I am skeptical. 92% accuracy ?????
How is the sensitivity and specificity calculated without any test data collected on real cases?
CT scan testing is not a benign/ easy modality. It is not going to be readily available plus it does exposes you to a lot of radiation.
If the device is just looking for an ARDS like picture and calling it Corona virus related presentation then surely in today's pandemic overwhelming cases with those CT scan features would be secondary to corona virus, however in a patient with ARDS secondary to any other Viral or bacterial etiology the false positive rate will be 100%(in the absence of blood or nasopharyngeal swab testing for actual presence of virus) and in the interim may lead to wrong diagnosis and treatment.

SiR you're absolutely right and under normal circumstances, your points would be right on the mark. However, given the very presence of an endemic viral agent, by rapidly scanning for signs of ards together with the right clinical history, fever, cough etc It is surely enough to make a reasonable educated guess and put that patient rapidly into the "presumed covid +ve" group for further triage. The point is to avoid releasing a positive infectee back into the environment. The risk of the consequences of that outweighs the risk of overtreating a few negative patients who have ards due to some other agent. Once isolated, the pcr test can still be done if there is doubt.

"Same is the case with false negative results. No pulmonary involvement with milder cases would have a high false Negative rate on such kind of testing with patient returning to general population and still spreading infection through contact."

Yes but the impact of the false negatives could be mitigated by enforced self isolation for two weeks or even quarantine in a special facility for those who are negative on CT. If they recover from their mild illness, fine. If they worsen, then do the PCR test.

 

[DocEinstein]

Big cities have lots of CT scanners in major hospitals. Repurpose them for rapid throughput. I am only saying this as this is what worked in China. It may not be doable in Pakistan but would be worth trying.

Countries that wait for the tsunami to hit and are reactive in their approach are being hit hardest. Being proactive and literally hunting for the diagnosis among anyone symptomatic - certainly in Karachi it is doable - is the best way forwards. This way we can isolate the patients immediately and start treating them, while keeping them from infecting others. Surely the Wuhan model must be applied to Karachi post haste.

S

SiR you're absolutely right and under normal circumstances, your points would be right on the mark. However, given the very presence of an endemic viral agent, by rapidly scanning for signs of ards together with the right clinical history, fever, cough etc It is surely enough to make a reasonable educated guess and put that patient rapidly into the "presumed covid +ve" group for further triage. The point is to avoid releasing a positive infectee back into the environment. The risk of the consequences of that outweighs the risk of overtreating a few negative patients who have ards due to some other agent. Once isolated, the pcr test can still be done if there is doubt.

"Same is the case with false negative results. No pulmonary involvement with milder cases would have a high false Negative rate on such kind of testing with patient returning to general population and still spreading infection through contact."

Yes but the impact of the false negatives could be mitigated by enforced self isolation for two weeks or even quarantine in a special facility for those who are negative on CT. If they recover from their mild illness, fine. If they worsen, then do the PCR test.

I agree with your points, however CT scans are a luxury available to Western developed nations readily (some not all).
I am not sure with the current situation in Pakistan about availability of CT scanning, but if we are just trying to look for pulmonary findings then any Physician worth his/her salt and a decent radiologist can easily pinpoint findings like increased bronchovascular markings even on a plain chest X-Ray and isolate those patients for further testing in the right clinical scenario.
At most the CT may provide some indications a few hours earlier, however isolation of such patients should have been started based on clinical picture and symptoms anyway.
And to make it clear further : Any patient who is developing ARDS to be noted on CT scan, will not be asymptomatic. His symptoms like shortness of breath would have started even before the typical radiological findings on CT scan or CXR seeking medical attention. (Radiological findings will always lag behind onset of symptoms and resolution of disease/radiological findings will always lag behind even after disease has cleared. For e.g a case of sever bronchitis or pneumonia will continue to have radiology findings on CXR/CT scan even 2 weeks after complete resolution of disease). Any such patient with symptoms of cough/ shortness of breath/fever today (whether history of exposure to corona virus or not ) is being isolated and definitive testing performed regardless, worldwide.

 

[khail007]

Countries that wait for the tsunami to hit and are reactive in their approach are being hit hardest.

Pakistan always falls in reactive category and that is also by dragging feet, to be proactive not in our DNA.
Further, CT-Scan is expensive for most of the Pakistanis ... to have that device in action and break the continuity of corona chain is very critical, which will be a positive aspect to fight against corona. , we have to increase the CT-Scanners at war footings.

 

[TNT]

Everyone is kind of jumping on the corona virus bandwagon but they miss the point. The cost of CT scan and PCR is same so why not PCR which is way more reliable and accurate. Apart from the cost aspect, CT scan facilities are limited and equipment expensive as compared to PCR. These students most probably trained tenser flow algorithm to detect virus in lungs CT scan, something not so novel or difficult.

 

[BATMAN]

Why than PM is not interested to help such talent?

Everyone is kind of jumping on the corona virus bandwagon but they miss the point. The cost of CT scan and PCR is same so why not PCR which is way more reliable and accurate. Apart from the cost aspect, CT scan facilities are limited and equipment expensive as compared to PCR. These students most probably trained tenser flow algorithm to detect virus in lungs CT scan, something not so novel or difficult.

As per my understanding this is fast test. Other test require many hours and is as reliable as these kids are claiming.

 

[masterchief_mirza]

Everyone is kind of jumping on the corona virus bandwagon but they miss the point. The cost of CT scan and PCR is same so why not PCR which is way more reliable and accurate. Apart from the cost aspect, CT scan facilities are limited and equipment expensive as compared to PCR. These students most probably trained tenser flow algorithm to detect virus in lungs CT scan, something not so novel or difficult.

I believe the CT approach could be much quicker in terms of results. Not sure about cost effectiveness but I would have thought that after the capital purchase is made, running CT scans is cheaper than PCR tests (remember transport costs of samples to the central labs too).

 

[RescueRanger]

Pakistan needs to make it's mind up which pathway for testing it needs to follow PCR, ImgH, RTK... This useless delay is playing with peoples lives. Problem is health advisors all bickering with eachother. S.Korea already offered antibody test kit to Pak...

 

[Defense Reader]

There was a program launched by shehbaz sherif he provides ct scan machines on distrct head quarter hospital level in punjab province.

 

[TNT]

I believe the CT approach could be much quicker in terms of results. Not sure about cost effectiveness but I would have thought that after the capital purchase is made, running CT scans is cheaper than PCR tests (remember transport costs of samples to the central labs too).

Latest research does point to CT as reliable and quick, though the capital required would be large. With ct scan i guess AI is not even needed to detect it.

 

[-blitzkrieg-]

the good thing about this is you dont need an expert to read CT scans..

Can CT scan detect early stage cases?

 

[DocEinstein]

the good thing about this is you dont need an expert to read CT scans..

Can CT scan detect early stage cases?

You DO need a qualified radiologist to read CT scan.
And NO, a CT scan will not detect early stage disease but a PCR assay will.

There was a program launched by shehbaz sherif he provides ct scan machines on distrct head quarter hospital level in punjab province.

Showbaz will obviously provide CT scan machines with a fat tooi commission for himself.