First DSI-Lawson Centre fellow bridges nutrition and AI to improve care for children

With experience in clinical nutrition and data sciences, Paraskevi Massara is drawing from both fields to improve personalized nutrition recommendations for children using AI. Her research — supported in part through the inaugural Data Sciences Institute-Lawson Centre Postdoctoral Fellowship for Data Science in Child Nutrition — incorporates large data sets to identify associations between nutrition, the gut microbiome and health outcomes.

“The AI revolution is touching every aspect of everyday life, including science and clinical practice,” says Massara, a postdoctoral fellow in nutritional sciences at the University of Toronto’s Temerty Faculty of Medicine and in the Child Health Evaluative Sciences program at The Hospital for Sick Children. “I became interested in using the new technologies that are available to us now to explore new aspects of nutrition science.”

Massara studied human nutrition and clinical dietetics in Athens, Greece, practising as a dietician for a short time before returning to graduate school for a master’s degree in data science and machine learning.

In 2016, she began a PhD at U of T, co-supervised by Temerty Medicine Professors Elena Comelli and Robert Bandsma, both researchers with the Joannah & Brian Lawson Centre for Child Nutrition.

Comelli’s lab investigates the relationship between diet and the gut microbiome, particularly in children, and how this relationship can impact long-term health. Massara was interested in this line of research, particularly with the opportunity to access omics data — large data sets for certain types of biological molecules, such as gut microbes — and cohort databases such as TARGet Kids! from Canada and the Pelotas Birth Cohort from Brazil.

Catherine Birken, a clinician-scientist at SickKids, professor at Temerty Medicine and researcher with the Lawson Centre for Child Nutrition, facilitated access to the Canadian TARGet Kids! cohort — enabling integration of rich clinical and longitudinal data into Massara’s research.

In her PhD project, Massara developed new machine learning approaches to study growth in children using these large data sets, mainly focused on improving study methodologies. For example, in one paper she compared the variation in growth patterns in children over time using different computer modelling techniques, and in another she examined how computer models treat outlier data in the detection of growth patterns.

Massara’s postdoctoral fellowship, with Comelli as primary supervisor and Birken and Charles Keown-Stoneman as co-supervisors, is a continuation of this work to bring together omics, clinical and diet data, and advanced AI technologies in nutritional science.

She says that by using machine learning and computer modelling to detect associations between gut microbiome, metabolic, clinical and growth data, researchers can gain insight and ultimately develop tailored nutrition recommendations based on a specific child’s characteristics.

“Our goal is to use all these various types of data to understand individual-level associations between diet, the microbiome and cardiometabolic health in children, which would allow clinicians to understand the needs of children and personalize recommendations,” says Massara.

The DSI-Lawson Centre fellowship, which is worth up to $70,000, is meant to accelerate the impact of data sciences in child nutrition. Massara says the fellowship has provided critical financial support and allowed her to make connections within the Data Sciences Institute, which will help in her future career working at the intersection of AI and nutrition.

“AI is giving us new opportunities to do more advanced tasks and study much larger amounts of data than we could in the past — including genetic and microbiome data, and even data from wearable devices. But it also has challenges, including that we don’t necessarily know how to use all this data effectively,” says Massara. “We need people with advanced knowledge of nutrition, data science and AI to make a good use of the data to improve real-world recommendations and care.”