New Delhi, May 16: Computer science engineers have set out to design an application that they say could help people conscious about calories and sugar pick the best ice cream to match their palates' desires with their nutrition concerns.

The researchers have proposed an application that they say could be used by consumers to select, and by restaurants or nutrition advisers to suggest, the best-matching ice cream flavours or combinations of scoops, taking into account their ingredients.

"This is a step towards helping people tailor their decisions about ice creams according to tastes and health needs," said Preeti Mulay, a software engineer and assistant professor at the Symbiosis Institute of Technology, Pune, who is leading the research.

However, doctors caution that the human body is too complex and such a software application may have limited practical use, particularly when managing diabetes.

Mulay and her colleagues have used public domain information on the constituents of three brands of ice creams - a butter-pecan flavour, and two vanilla flavours - to develop a small database of the levels of sugar, cholesterol, protein and dietary fibre in each ice cream.

Working with this database of only three ice creams, they used a well-known computational tool called analytical hierarchy process (AHP) to design a prototype application that they say could be used to recommend a specific type of ice cream to patients with diabetes.

Their results suggest that butter-pecan is richer in sugar than vanilla and should thus be avoided by patients with sugar concerns. In a simulated exercise that the engineers say has not been verified by doctors, the application results suggest that a patient with a blood sugar level of 223mg/dl may consume the vanilla flavours with 13mg/dl sugar, while a patient with a blood sugar level of 193mg/dl can consume butter pecan with 18mg/dl sugar.

The researchers say a database with nutritional features of many more ice creams will need to be established for the application to have any practical use. "This will become important because of the diversity of the ice cream flavours available," Mulay said.

Mulay and her colleagues have proposed the application in the journal Procedia Computer Science.

Nutrition experts say a tool to select appropriate ice creams may help people stay away from varieties with the highest sugar levels. But they caution that ice creams with relatively lower sugar levels should not be viewed as acceptable for those trying to curb sugar intake.

"People may use an application like this to quickly recognise what they should completely avoid," said Seema Gulati, head of nutrition research at the Fortis Centre for Diabetes Obesity and Cholesterol, New Delhi, who is not associated with the research in Pune.

"But even with relatively low sugar or calories, ice cream is still junk food, nothing changes that," Gulati told The Telegraph. "The calorie content of ice cream is disproportionately high in relation to its nutritional content."

Doctors point out that blood sugar levels are dynamic, constantly changing, and multiple organs and tissues in the body determine sugar levels.

"The pancreas, the liver, the intestine, muscles, fat, the brain - all of these influence sugar levels," said Nihal Thomas, professor and head of endocrinology at the Christian Medical College, Vellore.

Any software application, he said, may need to incorporate many more patient-related variables to be really effective. But it is hard to capture some of these variables in a quantitative manner, Thomas added.

Mulay says she plans to consult doctors and nutrition specialists in the coming months to incorporate their expertise before building the next version of the application. While the current version is designed for desktop computers, Mulay said, the long-term goal is to develop the application as a mobile app - an ice cream selection tool in people's pockets.

Computer scientists say the research effort represents a domain-specific use of analytical hierarchy processing. "This is a generic tool used in problems that involve multiple factors, where some are more important than others," said Pradip Ray, professor of industrial engineering at the Indian Institute of Technology, Kharagpur. "But such applications have to be developed in consultations with domain experts - in this case, doctors."