AI, pill boxes with reminders to tame TB

India has the highest burden of the infection, with 2.6 million patients accounting for a fourth of all cases across the globe. Yet, the country has set an ambitious target of eliminating the bacterial infection by 2025, five years ahead of the global Sustainable Development Goal target of 2030. To achieve this goal, the government is looking at several strategies, including detecting latent TB in close contacts of a patient and treating them, analysing whether vaccinating them can prevent the spread of the disease, looking for shorter courses of treatment, and involving communities to help spread awareness.

Medicine box with reminders

Several organisations have now come up with cardboard boxes that are enabled with a SIM card to remind patients to take their pills every day and send a message to their healthcare provider when they have.

The intervention was designed to ensure adherence to the treatment protocol with daily visits to DOTS centres. DOTS or Directly Observed Treatment-Short Course is a TB control strategy where the patients have to go to their assigned healthcare centres and take their medicines in front of the health workers. When they fail to do so, the health workers are required to visit their homes and motivate them to continue their treatment.

With TB patients being required to take their pills for six months to a couple of years, medicine fatigue and people giving up midway is well documented.

Scientists from NIRT conducted a trial of the medicine boxes, referred to as Medication Event Reminder Monitor (MERM), in patients with multi-drug resistant TB in Chennai and Mumbai. “The trial was to essentially see whether the use of the box was acceptable to the patients and whether it actually led to proper adherence to the treatment course even without the supervision of healthcare workers. Through interviews with the participants, healthcare workers, and analysis of the quantity of medicine in the urine of the patients, we found that there was a 70 to 80 per cent adherence,” said Dr N Karikalan, head of the department of social and behavioural research at ICMR-NIRT.

The field trial found that most patients appreciated the need for fewer visits to their healthcare centres only to collect the medicines and get tested, said Dr Karikalan. The internal partitions also helped the patients in organising and storing multiple drugs needed for the treatment of multi-drug resistant TB. “However, those with small one-room houses did not want to keep the huge boxes for fear that others would come to know that they had TB,” he said.

Dr Rajendra Joshi, deputy Director-General, National TB Elimination Programme, said innovations such as MERM could help in urban settings where many people do not allow ASHA workers to come to their homes due to the stigma associated with TB.

AI tool for detection

With thousands of X-rays collected as part of clinical trials and the national active case-finding survey, scientists from NIRT are now trying to create digital copies and train AI tools to identify normal and abnormal X-rays.

“We will build on the AI in three stages – in stage one, we will teach the algorithm to distinguish between normal and abnormal chest X-rays; in stage two, we will teach it to identify what the abnormality is; and hopefully, in stage three, we will be able to teach it to identify TB as opposed to say a malignancy in the lung in an X-ray,” said Dr G Narendran, who heads the clinical team at NIRT.

“We have been using the normal X-rays taken during the active case-finding survey and the abnormal ones from our archive to train the AI. The difficulty lies in teaching the algorithm the differences in physiology of people that are not necessarily pathological. We are currently validating the first phase where the AI can tell a normal and an abnormal chest x-ray apart,” he added.

This itself will have applications in community screening. As Dr Narendran explained, “Say for example we visit a locality and do 500 X-rays, a national prevalence of 316 TB cases per lakh population would mean we would find one or two TB cases. If the AI can tell a handful of abnormal X-rays apart, the doctor will need to check only those instead of going through all the 500 X-rays.”

The validation would require that the AI tools scan at least 5,000 X-rays – 4.900 normal ones and about 100 abnormal ones. “Developing an AI is like driving. The more you drive, the more confident you get and the fewer errors happen. Once you have driven 20,000 km, you know you are an expert. Once we finish the validation process, it can be tested in the community,” said Dr Narendran.

The second stage would require that the AI tool be trained in different types of abnormalities such as cavities or pleural effusions (build-up of fluid in the chest). This will bring the technology from the community stage to say a district hospital set-up. And, in the final stage the researchers hope to train it in not only identifying the abnormalities but telling whether they indicate it is TB or other lung diseases such as cancers.

“For this, we will have to reach out to tertiary care centres to seek X-rays of other abnormalities. However, it will be a very difficult process and we are just hopeful it happens,” Dr Narendran said.

BCG vaccination

With the Bacillus Calmette–Guérin (BCG) vaccine still a part of routine immunisation in the country, researchers from NIRT are looking at whether a booster shot in children between the ages of six and 18 from households where one person has TB can prevent the spread of the infection.

The institute will enroll 9,200 children from seven sites across the country to see whether re-vaccination can help in reducing the incidence of TB in the children as compared to the oral treatment that is currently offered to household contacts as part of the national elimination programme.

Although the BCG vaccination at birth is known to be effective in preventing meningitis (inflammation of brain membrane) and disseminated tuberculosis, it is not very effective against getting an infection, reactivation of a latent infection, and against the most common form of TB in the lungs.

After recruitment, the children will be tracked for two years to see how many develop active TB in the two groups – one that has received the vaccine and one that has received the oral medication. “The second shot will almost act as a booster in the children who have received their first dose as part of routine immunisation. We will see whether it prevents the infection. We know that most children develop active TB when they start going to school, college or when they start working,” said Dr C Padmapriyadarsini, director, NIRT.