A TU/Delft researcher has developed an AI coach to help people quit smoking and vaping. Nele Albers opted for so-called reinforcement learning (RL) when developing the tool. This is a form of machine learning where a model learns through rewards, similar to how humans learn behaviour. Albers based her research on insights from behaviour change theories and data from three large-scale studies with more than 500 participants each. This week, she defended her thesis.
Apart from the technical side, Albers also researched ethical, economic and psychological aspects. Moreover, she analysed how different factors contribute to effective behavioural change. These include issues such as how to convince smokers to quit, what is required of them and who should support them.
Potential of AI coaches
With her research, Albers has shown that AI coaches who take psychological principles into account have great potential to provide effective support to people who want to quit smoking. ‘My research shows how personalising support - by taking into account both a person's current and future state - increases the effectiveness of AI-based eHealth applications. This offers great potential for behaviour change,’ Albers said.
Albers also conducted research on both the algorithmic side and the interaction between smokers and AI coaches, with the aim of improving their effectiveness in supporting people who want to quit. One advantage of the AI coach she developed is that it is able to determine when it is better to call in a human coach as well. Finally, she also examined the tension between smokers' preferences and the advice of health experts. Health experts determine what is right, but smokers often view this differently. Albers' algorithm tries to strike a balance between the two perspectives.
For effectiveness, the research also shows, it is important that coaching is tailored to an individual's situation, such as when suggesting different activities. For example, the AI coach can encourage someone to think about stimuli that trigger the desire to smoke or the person they want to be in the future.
eHealth and AI coaching
Several digital tools have already been developed in recent years to help people quit smoking. This increasingly includes the smartphone. A gadget that just about everyone uses every day. A US study late last year showed that combining remote help and coaching with ‘physical’ help works particularly well.
Quitting smoking is hard enough for those struggling with this addiction. Help with quit attempts has been an issue for many (decades). Coaches, buddies, websites, leaflets, patches and, for some years now, remote eHealth and AI coaching. How successful these interventions and forms of support are depends largely on the smoker's will to quit and change his or her behaviour. A lot of research has been done on this too. The risk of relapse is always lurking for many smokers. This relapse, as well as lack of commitment, are the challenges facing AI coaches in eHealth applications. While many healthcare professionals endorse that these can effectively help bring about behavioural changes, their deployment is still disappointing.
