Leveraging AI for health behaviour change
Published on 27 June 2024
An app to fight obesity? Machine learning can help forecast and deter dietary triggers.
To Dr Jocelyn Chew, obesity is more than a lifestyle problem. Instead, she sees it as a chronic disease that requires a concerted amount of clinical and community support to manage. “It is not a condition where you can be discharged after surgery or after being prescribed with medication. It is a long-term condition that requires consistent monitoring and self-regulation,” she said.
Evidence is also clear, she noted, that the burden of cardiometabolic diseases such as stroke, diabetes and liver disease, will be alleviated by addressing the upstream problem of excess adiposity (body fat). “You solve this problem, you solve your DHL — diabetes, hypertension, hyperlipidemia (an elevated level of lipids like cholesterol in the blood),” she added.
To ease the burden of cardiometabolic diseases, Dr Chew has been working on an app called the Eating Trigger-Response Inhibition Program, or eTRIP. Unlike other weight-loss apps, what makes the eTRIP unique is that it is designed to predictively tackle a person’s dietary triggers in order to help them identify moments they could be in danger of overeating. Some examples of dietary triggers include emotional eating, a “4pm snacking time”, or even habits such as a family over-ordering during frequent gatherings.
A nurse by training, Dr Chew’s passion for research saw her pursuing a PhD at the Chinese University of Hong Kong, to become the youngest Nursing PhD holder in Singapore. The Assistant Professor at the NUS Alice Lee Centre for Nursing Studies, part of NUS Yong Loo Lin School of Medicine (NUS Medicine), also became the first nurse to receive training in the Stanford Biodesign methodology as a Singapore Biodesign Innovation Fellow from 2023 to 2024.
Achievements aside, she stays true to her priority of improving population health. Her philosophy underscores this. “While working in the hospital allows me to touch lives one patient at a time, research has the power to impact thousands simultaneously. By identifying trends and developing evidence-based interventions, we can implement widespread changes that improve health outcomes on a large scale, creating a healthier future for entire communities,” she shared.
Leveraging her special interest in driving innovation and data translation, artificial intelligence (AI), digital health and implementation science, Dr Chew serves primarily as a researcher within the National University Healthcare System (NUHS), and an educator at NUS Medicine and NUS Nursing.
Preventive and predictive: Using AI to identify and tackle dietary triggers
Dr Chew’s eTRIP app was designed after multiple exploratory studies to find out patients’ dietary triggers, identifying sustainable “baby steps” towards change and implementing “consistent health coaching” over a 12-week period (the typical timeframe for behaviour modification), instead of simply spewing general advice and guidelines on nutrition and exercise.
“During the app onboarding, users will engage in activities to elucidate their values, motivation, weight-loss goals and action plans. They will also be guided in how to formulate SMART goals and self-regulation strategies,” explained Dr Chew.
With this information, eTRIP then uses data analytics and AI to formulate personalised ‘just-in-time’ nudges, such as alerting the patient five minutes prior to their ‘snacking time’, logging meals via photos, and reinforcing healthy behavioural changes through repetition. There is also a large language model-based chatbot to provide precise encouragement and advice, and modules to help users reflect and self-regulate.
Studies have also been encouraging, she shared. Participants saw significant improvements by over 50 per cent in overeating habits, eating self-regulation, physical activity, depression and anxiety.
Shaping tomorrow’s health, today
Dr Chew has big plans for eTRIP. The next iteration of testing will incorporate biofeedback — glucose monitoring, weight monitoring and even ketones — to monitor the state of fat burn, she revealed. Eventually, she plans to incorporate the app into a nurse-led weight triage and management service, which she is also developing.
“Nurses have a strong role to play as they are there 24/7 with patients,” she elaborated. “There are a lot of opportunities in the inpatient setting for patients to learn healthy lifestyle habits. This, coupled with nurses’ foundational knowledge in managing a lot of chronic diseases, makes them the perfect health coach to follow through.”
Also in development to be used in tandem with eTRIP is Adipoview, an AI-based dashboard that Dr Chew has created with the team from the NUHS Group Chief Technology Office. Adipoview will gather electronic medical record (EMR) data throughout NUHS to generate data insights, such as the diagnosis rate of obesity and treatment rate, as a proactive way to predict who may need further support.
On the topic of predictive, precise and personalised medicine, Dr Chew believes that “AI is the cornerstone of preventive medicine and health behaviour change. Everyone has unique behaviours, needs and preferences; there's no way traditional healthcare can cope with this form of complexity in providing personalised and precision care,” she said.
In consultation with Dr Jocelyn Chew, Assistant Professor, NUS Nursing, NUS Medicine.
