To make a Covid-19 diagnosis, we have all had to go through this at some point or another: mucus is removed from the nose or throat with a cotton swab. This is then examined with a PCR test in a laboratory for the presence of the coronavirus. However, there are also many places worldwide where no laboratories are available, or where access to the well-known high-tech PCR test is impossible or more difficult. The use of AI, or deep learning, models to improve the diagnosis of Covid-19 using X-rays has already been investigated by several scientists worldwide in recent years.
AI program for Covid-19 diagnosis
The AI program that has now been described by Keelin Murphy and Bram van Ginneken of Radboud university medical center, among others, appears to be a good alternative to the antigen rapid test. The program combines an X-ray of the lungs with Point Of Care (POC) blood tests. Because most people who are tested for Covid-19 only experience mild complaints, lung x-rays rarely reveal pneumonia. However, in combination with one or more blood tests, the AI program does become a reliable test.
With the so-called Point Of Care tests, the result of what the doctor wants to measure is immediately displayed, without the blood having to be analyzed in a laboratory. One of the POC tests used measures the numbers of different white blood cells. An increase indicates an infection or inflammation. The other POC test measures C-reactive protein (CRP). An increased value of CRP is also an indication of inflammation. “If the X-ray data are combined with WBC, possibly supplemented with CRP, a diagnosis is created that is more sensitive than an antigen rapid test in the tested African regions,” the researchers said.
More accurate than antigen rapid test
The AI program has been tested at four locations in Africa. Locations where PCR tests are not available due to a lack of people and resources, but where X-rays can be taken and POC tests are available. More than 3,500 people participated in the tests. In addition to the accessible X-ray and POC testing, a PCR test was also carried out for all participants as a check.
The results of the AI program for Covid-19 diagnoses were very satisfactory. Compared to the rapid, also accessible, antigen rapid test, diagnosing Covid-19 using the AI program turned out to be more sensitive. This means that the AI program made more correct positive diagnoses.
“This is the first study to validate AI methods for COVID-19 detection in an African setting. It shows that screening for COVID-19 using AI and point-of-care blood tests is feasible and may be more sensitive than antigen tests,” the researchers write. In addition to diagnosing Covid-19, researchers note that AI programs that combine an X-ray with POC blood tests may also eventually be suitable for tuberculosis diagnoses in medically less well-resourced regions.
The researchers’ findings are published in Scientific Reports: COVID-19 screening in low resource settings using artificial intelligence for chest radiographs and point-of-care blood tests – Keelin Murphy, Josephine Muhairwe, Steven Schalekamp, Bram van Ginneken, Irene Ayakaka, Kamele Mashaete, Bulemba Katende, Alastair van Heerde, Shannon Bosman, Thandanani Madonsela, Lucia Gonzalez Fernandez, Aita Signorell, Moniek Bresser, Klaus Reither & Tracy R. Glass.
2023-12-04 05:58:47
#program #diagnose #Covid19 #ICThealth #official #knowledge #platform #healthcare #innovation
