George Salloum and His Team Want to Help People Get Healthy
Computer Engineering grad student creates software to take the mystery out of nutrition.
George Salloum, who is pursuing an MS in Computer Engineering at LAU, has developed an innovative software called Personalized Intelligent Nutrition (PIN), which is capable of using human-like logic to generate health and nutrition recommendations and suggest meal plans that account for food diversity and taste preferences. In addition, the platform includes a feature for monitoring progress and making adjustments. The application has shown through testing to be highly compatible with a human expert’s nutritional assessments.
Salloum, 24, has been working on this project since he was an undergraduate at LAU. He first formulated the idea – which combines his skills in computer engineering and interest in health and fitness – in his capstone project on which he collaborated with fellow student, Elie Semaan (BE ‘17). When Salloum enrolled in the master’s program, he decided to make PIN the topic of his thesis and was encouraged to pursue his interest by Assistant Professor and Interim Assistant Dean at the School of Engineering Joe Tekli, who had supervised Salloum’s capstone project.
Before deciding on the features of the PIN software, Salloum extensively researched the existing mobile solutions that monitor food consumption and offer meal plans. What he found was that the available tools have three major drawbacks. First, they only take into consideration a person’s height to recommend an ideal body weight, relying on the widely used Body Mass Index (BMI) as a scale. According to Salloum, this approach overlooks the user’s body composition in terms of fat and muscle content, which can only be assessed through the Body Fat Percentage (BFP). Second, the other applications have limitations when it comes to assessing food preferences and generating meal suggestions. Third, no tool has the ability to monitor progress and amend meal offerings midway through the plan to help users reach their target.
“We looked at what is out there, highlighted what was missing, and found that there is room for improvement,” said Salloum. “The solution is not always reducing caloric intake. Exercise is a must for people who are mainly sedentary. This is not just a mathematical calculation; we need to apply some logic to it.”
The PIN approach, which addresses the above shortcomings and more, consists of four simple steps. First, users input details about their gender, age, height, weight and BFP. Using the “fuzzy logic paradigm” which mimics human decision-making, the program processes the information to make recommendations about weight and BFP. Second, users provide information about their activity level and exercise preferences to receive recommendations about their target caloric intake and exercise options. The third step involves selecting foods and rating them according to preference, which PIN processes to generate meal plans. The last step is the software’s built-in ability to monitor the users’ progress and amend the caloric intake or exercise plan to ensure that targets are met.
The software’s meal planning process is based on a sophisticated paradigm that takes several factors into consideration and is one of PIN’s most distinguishing features. The tool ensures meal-food compatibility whereby each food item is matched with the suitable meal. For instance, if a person indicates that they desire fruit in their meal plan, PIN will slot it as a breakfast or snack option. The application also accounts for inter-food compatibility so that foods served within the same meal are complementary, such as rice and meat or cheese and bread. In addition, users’ ratings of their food preferences on a scale that ranges from “love” to “hate” will determine how often these ingredients appear in their plan. Another unique characteristic is that PIN keeps track of the daily food offerings to ensure diversity and variety.
The tool’s ability to simulate the human decision-making process was tested in an experiment using 25 male and 25 female subjects as well as 11 nutrition experts. After collecting personal data on each individual, PIN generated recommendations about weight, BFP, caloric intake, and exercise level. These results were then compared with the assessment of the human evaluators, and the similarity was close to 90 percent, a high correlation.
“There will never be a 100 percent match between results, even among human experts, as this process is not based on a single mathematical formula. There is some degree of common sense and experiential logic involved,” explained Salloum. “The similarity between our results and those of the experts was within the same range and sometimes surpassed the agreement between the experts themselves.”
In order to evaluate the meal plans, the researchers relied on expert as well as non-expert opinions. Nutritionists were asked to rate the meals on a scale from one to five to assess whether they met the compatibility and diversity criteria. The experiment also attempted to predict feedback from the general public by seeking the opinion of five non-experts who were also asked to rate meals based on their indicated preferences.
“The results were satisfactory, but some areas were better than others,” said Salloum.
Future development may provide an option to calculate BFP at the outset, as the present tool relies on the user to provide this information. Further refinement of the meal plan could take into account regional tastes, seasonal food items, and expiration dates.
Salloum credits LAU and its faculty, as well as external consultants, for their contributions to his research. He stresses Dr. Tekli’s role as an instructor and mentor who goes above and beyond duty to support his students and expose their work. Dr. Maya Bassil, associate professor of nutrition at the Department of Natural Sciences, too, was a major contributor to the research as her advice was critical during the experimental design stage. Another specialist, Eva-Maria Kahwaji who holds an MSC in Sports and Exercise Nutrition, also provided valuable advice to the research.
“Dr. Tekli never missed an opportunity to highlight the work at conferences and in publications. He is fair and rewarding, which is motivating to me,” said Salloum, adding, “The collaboration with Dr. Bassil also added a big validation to the work.”
Salloum’s research has been presented at several conferences and received with interest. A short version of the paper was published by the prestigious IEEE World Congress on Services at a conference in San Francisco last July. The work was also discussed at the IEEE Lebanon Communication Research Day held at the Order of Engineers in Tripoli in 2017, and at the Lebanon Diaspora Energy Conference in Beirut that same year.
Asked whether PIN could eventually make jobs in the field of diet and nutrition redundant, Salloum said that he did not see this development as a zero-sum game.
“I don’t think the tool will replace dietitians, but it will assist them in their jobs. They will use it to help them assess and speed up the meal-planning process. Many dietitians who have seen the software are excited about using it.”