In some of the world’s most remote and impoverished corners, where respiratory diseases abound and trained medical professionals fear to step in, the diagnosis is increasingly driven by artificial intelligence and the Internet.
In less than a minute, a new app on a phone or computer can scan an X-ray for signs of tuberculosis, Covid-19 and 27 other conditions.
TB, the deadliest infectious disease in the world, cost nearly 1.4 million lives last year. The app, called qXR, is one of many AI-based tools that have emerged over the last few years for screening and diagnosing TB.
“Among all applications of AI, I believe that digital interpretation of an image using an algorithm instead of a human radiologist is probably at the forefront,” said Madhukar Pai, director of the McGill International TB Center in Montreal.
Artificial intelligence cannot replace clinicians, warned Dr. Pie and other experts. But the combination of AI and clinical expertise is proving to be strong.
“The machine plus the clinician is better than the clinician, and it is also better than the machine alone,” said Dr. Eric Topol, director of the Scripps Research Translational Institute in San Diego and author of a book on the use of AI in medicine.
In India, where approx. a quarter of the world’s TB cases occur, there is an urgent need for an app that can mark the disease in remote locations.
Chinchpada Christian Hospital in Nandurbar, a small town in northwestern India, serves members of the Bhil tribal community, some of whom travel up to 125 miles to visit the center. The 50-bed hospital has eight doctors and only the most rudimentary medical equipment.
Simdega, one of the 20 poorest districts in India, is clear across the country and is isolated from the nearest town, Rourkela, with almost five hours of travel on bumpy roads. The tribal population of the district lives in small villages surrounded by dense, evergreen forest. Simdegas medical center, which has 60 beds and three doctors, is in a clearing of the forest – “literally in the middle of nowhere”, said director Dr. George Mathew.
The lean staff must manage everything that comes, “from malaria to heart attack to seizures to head injuries,” said Dr. Mathew. Over the years, he has taught himself to read X-rays, and when he gets stumped, he appeals to radiologists among his distant friends and former colleagues.
Although Nandurbar and Simdega are separated by more than 800 miles, their population is surprisingly similar. Malaria, sickle cell disease and tuberculosis are rampant among them, exacerbated by poverty, dependence on spiritual healers and alcoholism – even among children.
“TB tends to be neglected and the diagnosis is often delayed,” said Dr. Ashita Singh, Chief Medical Officer at Nandurbar Hospital. When people arrive at these medical centers, they are often “very, very ill and have never even been evaluated elsewhere,” she said.
However, in some patients, the radiographs show signs that are too subtle for a non-expert to detect. “It’s in the group of patients where AI technology can be of great benefit,” said Dr. Singh.
The arrival of coronavirus – and the ensuing lockdown – cut off these remote hospitals from the nearest cities and also from radiologists. It also delayed further and complicated TB diagnoses because both diseases affect the lungs.
A few months ago, both hospitals started using qXR, an app made by the Indian company Qure.ai and subsidized by the Indian government. The app allows the user to scan an X-ray. If it finds signs of TB, it assigns the patient a risk score. Physicians can then perform confirmatory tests on patients at the highest risk.
At the hospital in Nandurbar, the app helped diagnose TB in 20 patients in October, Dr. Singh.
Apps like qXR can also be useful places with a low incidence of TB and for routine screening of people with HIV who are at high risk of getting TB, as well as for those who have other conditions, experts say.
Confused by the terms of the coronavirus test? Let us help:
- Antibody: A protein produced by the immune system that can recognize and bind accurately to specific viruses, bacteria or other invaders.
- Antibody test / serology test: A test that detects antibodies specific for coronavirus. Antibodies begin to appear in the blood about a week after the coronavirus has infected the body. Because antibodies take so long to develop, an antibody test cannot reliably diagnose an ongoing infection. But it can identify people who have been exposed to coronavirus in the past.
- Antigen test: This test detects bits of coronavirus proteins called antigens. Antigen tests are fast and take as little as five minutes, but are less accurate than tests that detect genetic material from the virus.
- Coronavirus: Any virus that belongs to the Orthocoronavirinae family of viruses. Coronavirus that causes Covid-19 is known as SARS-CoV-2.
- Covid19: The disease caused by the new coronavirus. The name is an abbreviation for coronavirus disease 2019.
- Insulation and quarantine: Isolation is the separation of people who know they are sick with a contagious disease from those who are not sick. Quarantine refers to restricting the movement of people who have been exposed to a virus.
- Nasopharyngeal swab: A long, flexible stick with a soft cotton swab that is inserted deep into the nose to get samples from the space where the nasal cavity meets the throat. Samples for coronavirus tests can also be collected with swabs that do not go as deep into the nose – sometimes called nasal swabs – or oral or throat swabs.
- Polymerase Chain Reaction (PCR): Researchers use PCR to make millions of copies of genetic material in a sample. Tests that use PCR allow researchers to detect coronavirus even when it is scarce.
- Virus load: The amount of virus in a person’s body. In people infected with coronavirus, the viral load may peak before they begin to show symptoms, if symptoms appear at all.
“Most x-rays of the breast of people suspected of having tuberculosis are read by people who are not external experts in interpreting them,” said Dr. Richard E. Chaisson, a TB expert at Johns Hopkins University. “If there was an AI package that could read the x-rays and CT scans for you in a remote emergency room, it would be a big, big step forward.”
qXR is among the more promising of the AI-based apps for detecting TB. The company that made the app did not realize this potential until a doctor at an Indian hospital suggested it a few years ago.
In studies comparing different AI applications conducted by the Stop TB Partnership, all AI apps surpassed the experienced human readers, and qXR seemed to perform best.
The app identifies TB with an accuracy of 95 percent, according to Qure.ai’s CEO, Prashant Warier. But this level of precision is not based on real conditions, as Dr. Topol called “a common problem” with AI-based apps. A TB program may be less accurate in the United States or Western Europe than in India because the incidence of the disease is lower in these places, Dr. added. Topol.
The app has only been tested in adults, but it is now used for children 6 and over. X-rays of the chest are especially useful for pediatric tuberculosis because approx. 70 percent of cases in children can not be confirmed by laboratory tests, says Dr. Silvia S. Chiang, an expert in pediatric tuberculosis at Brown University.
“There is a huge shortage of trained professionals who are good at interpreting X-rays from the chest,” she said, “so the development and validation of computer-aided X-ray reading technology in children will greatly help.”
Qure.ai said it was testing its app on children in Bangladesh and that it would release the data early next year. Meanwhile, qXR and other apps are improving because they learn as they go.
“The more x-rays you feed the beast, the better it gets,” said Dr. Pai.
Experts were optimistic about all that AI-based apps could have a huge impact on TB control, especially in countries like India that lack medical resources.
“I just dream of a time when something like this would be available to all the small primary and secondary health centers in the public sector that are reluctant to do x-rays because they do not have the confidence to read them,” Dr. Singh said. “If this were to be made available to all X-ray centers in India, I think we could beat TB.”