New Delhi: Healthcare used to be all about doctors, drugs and diagnostics once. In the new, post-pandemic world around us now, a new element has entered this simple equation — artificial intelligence (AI).
The technology, which is now a big part of several sectors across industries, is being used in several hospitals throughout India for a variety of purposes — from reading scans to predicting risks.
Covid-19 has only firmed up AI’s role in healthcare, with hospitals even detecting the extent of lung damage, especially during the second wave when the Delta variant of the SARS-CoV-2 virus led to rapid deterioration of pulmonary health.
While the use of AI still remains widespread in radiology, particularly for diseases like tuberculosis (TB), it is increasingly being used for other purposes, including as a part of preventive health checks, particularly to predict cardiac disease risk.
The government of India too is betting big on AI to help it keep track of disease outbreaks across India. A tool currently being developed by a private company along with the National Centre for Disease Control (NCDC) will scan all media reports related to health, to create a database of outbreaks of 33 diseases — some with the potential to become epidemics — that are monitored under the Integrated Disease Surveillance Programme (IDSP).
For most hospitals and companies that are developing AI tools specifically for use in healthcare settings, the journey began about four years ago, mostly with radiology.
“The initial use of AI was in radiology. There is a feature we are all familiar with — face recognition. We started using AI for image identification and then tied up with companies to validate the findings. During the Covid pandemic, especially during the second wave, we used AI to analyse CT scans to find out the degree of lung infection,” said Dr Bharat Aggarwal, chief technology officer, Max Healthcare, Delhi.
“We are also doing a lot of research work on the use of AI for TB detection and have tied up with some healthcare start-ups and are also looking at the use of AI to detect psychiatric illnesses and cognitive disorders of the brain,” he said.
Also read: Two hospitals in Faridabad, Chennai launch new centre for fast organ transplant services
Boost for under-served areas
Seventeen years after the National Rural Health Mission was launched, infrastructure and manpower shortfalls continue to plague vast swathes of India, particularly the rural areas. This is where AI could become an enabler, companies working in the field say.
AI-based solutions company Deeptek.ai was set up in 2017. Among the partners at the firm are Indian Institute of Technology (IIT) graduate Ajit Patil and radiologist Dr Amit Kharat. The many projects that the company is working on include one in Chennai that uses AI for TB screening, Patil said.
“Our first such project is in Chennai. We are working with the Stop TB partnership to run these TB screening vans. Earlier detection rates were about 50 in 1,00,000 scans but now with the use of AI, it has gone up to 500. Now, we are taking the same solution to rural clinics in Uttar Pradesh,” he told ThePrint.
In the long run, he said, AI could become autonomous (like driverless cars) where the role of human experts would be limited to validation.
“This is important particularly for a country like India because there are just 12,000 radiologists for the entire country. There is a shortage of radiologists in the entire world in fact,” he added.
Prashant Warrier, CEO of qure.ai, another company working in the healthcare sector, told ThePrint that his firm is currently working at approximately 600 sites on projects that include TB screening, early lung cancer detection and also helping stroke patients get early interventions.
The company’s AI-based solution for neurological patients, called qER, can quantify critical abnormalities like brain bleeds, cranial fractures and stroke to help emergency care doctors prioritise conditions that need urgent attention.
Wadhwani AI, which is partnering with the government of India, is working on multiple TB projects that include not just radiological tools but also prediction of risk for ‘Loss to Follow Up’ (people who don’t turn up for follow-up appointments). This is important for TB as the treatment is a long process that could continue for 6-9 months and many people experience side effects that make them drop out midway.
“To assist frontline health workers identify underweight neonates and monitor their growth, we are developing a smartphone-based technology that provides accurate, timely, geo-tagged and tamper-proof weight estimation,” a spokesperson for Wadhwani AI told ThePrint.
“Globally, 2.4 million children died in the first month of life in 2019 — approximately 6,700 neonatal deaths every day. Research suggests that many of these deaths are preventable if the baby’s weight can be determined in the first week after birth — an enormous problem in countries such as India where a significant number of births still take place at home, with no trained midwife or medical professional in attendance,” the spokesperson said.
During the pandemic, the company had also developed an AI-based cough sound analysis technology to help identify at-risk Covid patients before administering lab-based tests, even if they were asymptomatic.
Also read: Move over Zomato’s 10-min delivery, Hyderabad platform aims to provide ambulance in 8 mins
There is, however, a long road ahead and challenges galore, said Dr Aggarwal.
“Systems are not fully attuned to the use of AI, even IT systems are not. The accuracy of communication alerts needs to be increased. Over time though, as the database gets broader, there will be higher accuracy. It has many advantages,” he said.
“It is machine learning, so there is no distraction, though the lack of lateral thinking can be both a positive and a negative. Unlike a human doctor, AI will not look beyond the dataset/images it is presented with,” he added.
The best known healthcare AI programme was IBM’s Watson, used for cancer detection. In a study published in the Future Healthcare Journal, US researchers wrote: “IBM’s Watson has received considerable attention in the media for its focus on precision medicine, particularly cancer diagnosis and treatment. Watson employs a combination of machine learning and NLP (natural language processing) capabilities.
“However, early enthusiasm for this application of the technology has faded as customers realised the difficulty of teaching Watson how to address particular types of cancer and of integrating Watson into care processes and systems. Watson is not a single product but a set of ‘cognitive services’ provided through application programming interfaces (APIs), including speech and language, vision, and machine learning-based data-analysis programs. Most observers feel that the Watson APIs are technically capable, but taking on cancer treatment was an overly ambitious objective.”
The primary challenge for AI is data collection, given the concerns about data security and data privacy.
“Algorithms used in machine learning systems and AI can only be as good as the data. High quality data is crucial for high quality algorithms. However, the call for high quality data in discussions around AI often remains without any further specifications and guidance as to what this actually means,” explained the Wadhwani AI spokesperson.
“Attributing to the fact that there are several sources of error in all data collections, the users of AI-related technology need to have a sound understanding of the source of the data and the potential shortcomings of the data. AI systems based on incomplete or biased data can lead to inaccurate outcomes that infringe on people’s fundamental rights, including discrimination. Being transparent about the data used in AI systems aids in preventing possible rights violations. This is especially important in times of big data, where the volume of data is sometimes valued over quality,” the spokesperson added.
(Edited by Amit Upadhyaya)
Also read: Inching towards self-reliance: India now makes 35 drug raw materials that were imported before