Can AI-backed tuberculosis software help detect lung cancer too?

Mumbai: In 2022, BMC health officials found that an artificial intelligence-backed software detected tuberculosis in some people months before the disease's tell-tale symptoms such as coughing emerged.

Now, Qure.ai, a city-based startup that fitted the AI software in nine BMC hospitals and a mobile van wants to update the programme to detect lung cancer as well.

BMC executive health officer Dr Mangala Gomare said, "Of the 1,050 people diagnosed with TB by AI in 2022, 35 per cent were asymptomatic. It was helpful that they were identified before they could spread the disease in the community."

"

Our AI-enabled chest X-ray solution for TB screening can also be used to screen for other abnormalities like lung nodules that could be indicative of cancer," Qure.ai founder Prashant Warier said, adding, "We'll explore a partnership with BMC to target lung cancer by expanding our programme in locations where AI-based TB screening is active."

City startup Qure.ai, which is helping the BMC with AI-backed software for early detection of TB cases, is exploring a tie-up with the civic body to target lung cancer too. BMC executive health officer Dr Mangala Gomare said it would be useful if the algorithm used for TB could also be used to detect lung cancer.

Talk on AI is dominated by chatbots such as ChatGPT, Bing or Bard, but senior endocrinologist Dr Anoop Misra believes AI is leading a "revolution" in healthcare. From picking up trends, analysing results and helping with research writing, AI is fast becoming a part of medicine.

As the editor of 'Diabetes & Metabolic Syndrome: Clinical Research & Reviews', an indexed medical journal, the Delhi-based Misra has already led and edited reviews on AI-based tools in medicine. "There is a huge potential with AI in every branch of medicine, be it radiology, ophthalmology or multiple clinical fields," he said, adding that physicians who cannot understand AI "will be left behind".

Across the world, multiple studies are underway to check if and how AI and Big Data can further help medicine. In Zydus Hospital in Ahmedabad and Aster MIMS in Kozhikode, Warier's AI-based stroke management programme is leading to quicker diagnosis of stroke. "Our apps help clinicians and Medtronic India is helping hospitals with training and solutions," he said.

Over a week back, a research paper in 'Nature Medicine' highlighted that an AI-based programme could detect pancreatic cancer, which is most difficult to catch in its early stages, at least three years in advance. The joint study by Harvard University and the University of Copenhagen used patients' medical records for the prediction.

But not all are impressed as yet. According to Dr Shailesh Shrikhande, deputy director at Tata Memorial Hospital and a pancreatic cancer surgeon, "AI currently depends on big data in various algorithms. If the data is of good quality, then we can get good inference and information," he said. However, accessing high quality data is not easy, especially when an algorithm needs tens of thousands of data points.

"If we have an AI tool based on high quality data, it would definitely serve as a complimentary guide for doctors," said Dr Shrikhande. At the moment, however, there is no pancreatic cancer AI-based algorithm that can be used for the general public. In their conclusion in the Nature Medicine article, the researchers said their results would help "design realistic surveillance programs for patients at elevated risk" of pancreatic cancer. "It cannot be used as a general screening tool as yet," he said.

Warrier, too, said that most concerns regarding AI in healthcare originate from the need for "unbiased" and "secure handling of patient data". He added, "AI is not a replacement to healthcare providers. Instead, it is an effective tool to augment their efforts."

Dr Misra concurred, "Doctors have to use their judgement and experience to interpret AI data at the moment. Nothing is 100 per cent yet."