By Dave DeFusco
Katz School and UMass Dartmouth researchers have introduced a novel approach to better understand the variables associated with weight loss by utilizing an advanced mathematical technique. Their findings, outlined in the white paper, “A Choquet-Integral Based Approach to Identify Weight Loss Component Subsets,” were presented in June at the IEEE/ACM international conference on Connected Health: Applications, Systems and Engineering Technologies (CHASE).
The Choquet Integral, commonly used in theoretical mathematics and economics, has seen limited application in the biomedical field; however, this innovative technique is now being applied to health data, showing a robust method to target and optimize the various factors and metrics that can influence or measure weight loss.
The new method pinpointed which health factors are most crucial for weight loss, offered a more precise and efficient way to study weight loss compared to traditional methods, and reduced errors and maximized useful information by focusing on the most relevant data. Unlike traditional measures that give definite values, like size or length, the Choquet Integral with a so-called fuzzy measure can handle uncertainty or overlap, providing a more flexible way to evaluate complex, interconnected data.
“We believe this methodology could pave the way for more efficient and accurate health data analysis, ultimately contributing to better health outcomes and advancing the fight against obesity,” said Matthew Fried, the first author of the paper and a Ph.D. student in mathematics under the supervision of Dr. Honggang Wang, a co-author of the paper and chair of the Graduate Department of Computer Science and Engineering at the Katz School.
The methodology was tested on four different datasets, including random numbers, manufactured data, standard heart data from the UC Irvine Libraries and National Institutes of Health (NIH) health data. The technique effectively distinguished between real data and noise, demonstrating its relevance in modeling interactive features, measuring features such as insulin and glucose levels; bad cholesterol (LDL); good cholesterol (HDL); height, and more.
“We studied how different health factors affect each other, whether positively or negatively, using this special mathematical approach,” said Wang. “This method helped us understand more clearly which biological factors are most important for weight loss.”
The application of the Choquet Integral to health data analysis represents an innovative cross-disciplinary approach, promising enhancements in machine learning models by selecting reduced versions of power sets, making the analysis more efficient and manageable and potentially transforming weight loss studies and other biomedical fields.
“The benefits of this technique extend beyond weight loss studies,” said Hua Fang, a co-author of the paper and professor of computer & information science at the University of Massachusetts at Dartmouth and UMass Chan Medical School. “It has broad potential applications in various biomedical fields where analyzing complex inter-variable relationships is crucial.”
