Risk-prediction tool for diabetes patients

A team of Indian scientists has set out to combine elements of disease biology, mathematics and software to predict for patients with diabetes the course of their disease and the risk of complications.

By G.S. Mudur
  • Published 7.04.16
  •  

New Delhi, April 6: A team of Indian scientists has set out to combine elements of disease biology, mathematics and software to predict for patients with diabetes the course of their disease and the risk of complications.

The scientists in a government laboratory and a private biotechnology company today signed a pact to develop an artificial intelligence platform that could be used to forewarn patients about the complications of diabetes involving the kidneys, the eyes or the cardiovascular system.

The research effort at the Drug Discovery and Research Centre (DDRC) of the Translational Health Science and Technology Institute in Haryana's Faridabad and the Hyderabad-based Revelations Biotech will seek to identify biomolecules in patients' bloodstreams that could serve as early warning signals to predict the progression of their disease. The Faridabad institute is a laboratory funded by the department of biotechnology under the science and technology ministry.

"Most of the complications of diabetes emerge over several years during the course of the disease," said Kanury Rao, an immunologist and head of the drug research centre at the institute. "We want to detect molecular signatures for different stages of the disease and understand how they change as complications emerge."

The scientists, hoping to release a prototype of their proposed artificial intelligence platform in about two years, say the product is intended to serve as a tool for doctors managing patients with diabetes.

Their project comes at a time India is experiencing a rapid rise in the number of diabetes patients, from an estimated 32 million in 2000 to about 63 million in 2013.

A study released today in the journal Lancet has estimated that between 1980 and 2014, the prevalence of diabetes in India increased from 3.7 per cent to 9.1 per cent among men, and 4.6 per cent to 8.3 per cent among women.

Doctors not associated with the project say a predictive artificial intelligence program would have the potential to help patients take precautions before symptoms of complications trouble them or before current laboratory tests can detect the complications.

Among the major complications of diabetes are kidney damage that can progress to life-threatening kidney failure and vision-threatening diabetic retinopathy, a condition caused by damage to the blood vessels in the retina.

"Among the earliest signs of diabetes-linked kidney damage is the secretion of protein in urine, but by the time this happens, the kidneys have already lost a significant proportion of their cells," said Oommen John, a physician and senior fellow at The George Institute for Global Health, a research institution in New Delhi. "A molecular marker to predict this complication ahead of this quantum of damage may allow time for preventive action."

The drug research centre team includes biologists and mathematicians who will analyse how the values of each of about 6,000 biomolecules in the bloodstreams of diabetes patients will change over time during the course of their disease.

The scientists, collaborating with doctors at the All India Institute of Medical Sciences, New Delhi, and Pune's King Edward Memorial Hospital, will build a database of the biomolecules studied in about 2,000 patients with different stages of diabetes. Some patients are in the pre-diabetic stage, others have just been diagnosed, while some have already developed the complications of various susceptible tissues or organs.

"To develop this artificial intelligence platform, we'll need to borrow ideas from mathematics and the theory of chaos that deals with how tiny fluctuations can make a huge difference over time," Rao said. "The goal is to develop a system that could provide a patient a probabilistic measure of his or her risk of developing a complication - for example, a typical output would predict a patient has a 90 per cent probability of developing diabetic kidney disease within two years," he said.

Revelations, the Hyderabad biotech firm, will contribute expertise in data mining and artificial intelligence, said Ravi Chandra Beeram, its managing director. The company also hopes to work on product development once a prototype emerges in about two years, a DDRC scientist said.