How AI is revolutionising cancer research at Indian educational institutions

As October, the breast cancer awareness month, approaches, Khalid Raza and his team at Jamia Millia Islamia remain hard at work. They are harnessing artificial intelligence (AI) for drug design to fight the global menace.

Raza, an associate professor of computer science, has received a grant worth Rs 94 lakh from the Indian Council of Medical Research (ICMR) for his three-year research, which includes exploring a drug compound already patented by him.

But his lab is just a part of a quiet revolution happening in India’s educational institutes, where AI is increasingly being used to advance cancer research.

Also read: Cancer deaths to rise by 75 pc in 25 years, ageing among factors: Study

AI for breast cancer research

“The idea was to use generative AI to generate some drug compounds and drug molecules. The drug molecules should be so precise that they would work on the target proteins of breast cancer," Raza told The Federal.

"There are many, many target proteins of breast cancer, and most of the drugs that are in the market also target similar proteins. But when we do it with the help of the AI, we have a chance of getting very high precision, which means an exact binding location of the protein where this drug has to bind,” he added. His patented drug compound is named ‘DdpMPyPEPhU’.

Many institutions are working on detection and diagnosis — an equally important part of cancer research. The IIT-M, for example, came up with ‘PIVOT’, an AI-based tool that uses gene data to identify cancer-causing genes, which they say will help in providing personalised cancer treatment strategies.

“So, the idea was to optimise the drug compound using some other AI methods and then synthesise it in the chemistry lab. AIIMS (All India Institute of Medical Sciences) is also a partner in this particular funded project. So, thereafter, they will test it on a mouse model that is called a PDX mouse model,” he said.

Also read: Enteromix explained: Can Russia’s cancer vaccine really save lives?

A PDX mouse model is one where a tissue from a cancerous tumour is transplanted into mice.

Disease mechanism

Raza has done work on how AI can be used for not only drug design, but also to understand the mechanism of disease.

“Mechanism means to see if there is any alteration at the gene expression or a genetic level, and how that alteration or mutation is triggering a disease. Basically, we have multiple biological pathways, and each pathway has some predefined function,” he said.

“So, when a disease occurs because of a mutation and some external perturbation, those pathways will not function according to their natural way. So, with the help of these methods, we can detect which pathways are being affected and how the disease is progressing. Understanding the disease mechanism is very important for solving it,” Raza added.

Also read: Can you get cancer by drinking hot tea or coffee? Here’s what evidence says

IIT-M initiative

Most other institutions are working on detection and diagnosis — an equally important part of cancer research. The Indian Institute of Technology, Madras (IIT-M), for example, came up with ‘PIVOT’, an AI-based tool that uses gene data to identify cancer-causing genes, which they say will ultimately help in providing personalised cancer-treatment strategies.

It has built prediction models for three cancers, including breast invasive carcinoma, colon adenocarcinoma, and lung adenocarcinoma.

The prediction is “based on a model that utilises information on mutations, expression of genes, and copy number variation in genes and perturbations in the biological network due to an altered gene expression”.

Also read: 60 pc of liver cancer cases preventable, but conditions apply: Medical gastroenterologist

The research was led by Prof Raghunathan Rengaswamy, dean (Global Engagement) and professor at the Department of Chemical Engineering; Karthik Raman, associate professor, and others, and was published in a peer-reviewed journal called Frontiers in Genetics.

Amity University’s dedicated AI lab is also doing pioneering work on cancer, making various aspects of cancer detection more affordable for the general public. Prof MK Dutta said in Tier-II or Tier-III cities, it is difficult to get facilities such as pap smears and mammograms and that too at affordable cost and they are trying to make the detections affordable.

“Although there are tools available to identify personalised cancer genes, they use unsupervised learning and predict based on the presence and absence of mutations in cancer-related genes. This study, however, is the first one to use supervised learning and takes into account the functional impact of mutations while making predictions,” as per a statement by IIT-M.

IIITH researchers build ML model

International Institute of Information Technology, Hyderabad (IIITH) researchers PK Vinod, CV Jawahar, and Sairam Tabibu have built a machine learning (ML) model to detect and subclassify renal cancer and also predict patient survival outcome using digital histopathological images. The research was published in the journal Nature.

Also read: BMI spike linked to breast cancer risk in post-menopausal women

The researchers first distinguished renal cancer tissue from normal tissue. The model then automatically classified the different renal cancer subtypes. From these subtype classifications, they identified features that could help predict patient survival.

The model achieved over 90 per cent accuracy in detecting whether a histopathological image showed a tumour and 94 per cent accuracy in classifying cancer subtypes. Survival predictions were based not just on nuclei features but also on tumour-shape features, which were found to be significant.

Amity's affordable cancer detection

Amity University’s dedicated AI lab — the Amity Centre for AI — is also doing pioneering work on cancer, making various aspects of cancer detection more affordable for the general public.

Prof MK Dutta is working with Professor Chandan Das from AIIMS on gallbladder carcinoma and how to use the ultrasound images of the gallbladder for the detection of carcinoma.

Also read: Can you prevent prostate cancer? What to know after Joe Biden's diagnosis

They’ve also worked on polycystic ovary carcinoma and skin cancer, and have written papers on some histopathological image-based cancer diagnostics.

The research has been published in peer-reviewed journals published or managed by Elsevier, such as Computers in Biology and Medicine and Computer Methods and Programs in Biomedicine.

Trained model

“In a country like India, if you go to tier-II or tier-III cities, you don't have facilities like pap smears and mammograms. And they are costly. So, what we are trying to do with artificial intelligence is to make these detections affordable,” Dutta told The Federal.

Also read: Idis can cause cancer: Karnataka bans plastic sheets for steaming humble food

“For example, in the case of breast cancer, we have trained a model where some of the demographics of the patient have to be entered. Demographic means age and some preliminary questions. And one blood test has to be done, which doesn’t cost much, just around Rs 150-200. This test has to be done to get some biomarker values.

"So, once you do this blood test and these values, you feed them in our AI model, in our artificial intelligence model, and it will predict if the person is healthy, or which grade of cancer it is,” he said, adding that the accuracy for the same was around 98 per cent. The centre has also done similar work on cervical cancer.