Innovation: How AI is helping to tackle TB in Tondo

Tuberculosis (TB) is the world’s deadliest infectious disease, disproportionately affecting the poorest communities. In a densely populated slum in the Philippines, MSF teams and local partners have been fighting the disease since 2022. Using machine-learning algorithms trained on large chest X-ray datasets, they've developed an innovative approach...

Overcrowded and confined living conditions, unsanitary housing, poor ventilation and high humidity create a breeding ground for the spread of TB bacteria, particularly among children, the elderly and people with weakened immune systems.

“Our experience conducting active TB case-finding in Tondo, one of the most overcrowded slums in the world, shows that around 5 percent of the population has confirmed TB disease in the district, which is nearly five times the national average of 1-1.5 percent,” explains Marve Duka, medical team leader for MSF in Tondo. 

The challenge of diagnosing TB in Tondo

But the onset of TB is gradual, making diagnosis complex.

“More than 50 percent of people diagnosed with TB in our program have no symptoms at all”, says Marve Duka.

As a result, many Tondo residents do not seek treatment until their symptoms have become severe, increasing the risk of transmission to others in the meantime.

Screening asymptomatic people, especially those who have been in close contact with a TB patient, is key to early diagnosis. Sputum—a type of thick mucus from a patient's lung—can be tested for the presence of the bacteria.

But collecting the sputum can be difficult, and when the level of bacteria is low it can give false negative results. This means it’s difficult to use for screening.

To overcome these challenges and increase our ability to detect TB, MSF is using a different set of tools. The teams go from neighbourhood to neighbourhood in a mobile X-ray truck.

Chest X-rays are commonly used to help diagnose pulmonary TB (TB on the lungs), but here the radiologists are assisted by a promising new tool based on artificial intelligence (AI). 

The role of AI in detecting TB

Computer-aided detection (CAD) is a tool that uses AI machine-learning algorithms trained on large chest X-ray datasets. CAD analyses chest X-rays for abnormalities suggestive of tuberculosis.

In Tondo, CAD technology has been implemented by MSF as a complementary tool to make TB screening faster and more efficient, by helping radiologists and doctors to identify early signs of TB disease.

CAD results can be processed in less than a minute, generating a TB score that helps clinicians determine whether a patient needs to be referred for further testing.

If a patient's CAD score exceeds a certain threshold, this indicates the possibility of tuberculosis, and the patient will have a sputum sample taken for confirmation by molecular tests such as GeneXpert, a diagnostic test that can identify Mycobacterium tuberculosis DNA and resistance to the antibiotic rifampicin.

Using CAD, MSF can screen between 120 and 150 patients a day. The sputum samples are taken to the laboratory the same day and results are generally available within two to three days.

Early detection leads to earlier treatment, which significantly reduces the risk of complications such as drug-resistant TB and organ damage and may decrease transmission of disease. For patients, this means a higher likelihood of recovery, fewer hospitalisations, and a better quality of life.

The implementation of CAD in Tondo in 2022 was accompanied by a study by Epicentre, MSF’s epidemiology and medical research arm, which showed that CAD technology was met with positive feedback from the community. 

A dedicated team to support patients with TB

MSF's comprehensive approach to TB care in Tondo goes beyond diagnosis and treatment. As soon as the diagnosis of TB is confirmed, a patient support team announces the diagnosis, provides advice and health education, ensures that patients understand what TB is and why adherence to treatment is essential, and that they receive the care they need.

“After just two weeks of treatment, patients are no longer contagious, which is very reassuring for patients and their relatives,” explains Belen Rance, a nurse who is part of the MSF patient support team.

The patient support team is also involved in contact-tracing activities, reaching out to TB patients’ household members to identify potential contact persons. Family members, particularly children, may be scheduled for chest X-rays and provided with TB preventative treatment if necessary.

TB treatment is provided free of charge for six months, ensuring that cost is not a barrier to accessing care.

However, logistical challenges persist. Some patients, especially those who are working or attending school, face difficulties attending regular appointments at health centres.

By bringing care closer to home, MSF is ensuring that more patients have sustainable access to TB care.