The technology developed at Lumme continuously analyzes sensor data from a wristband and a mobile phone to detect smoking and high-risk triggers for smoking lapses. This allows us to automatically detect triggers associated with each user’s smoking pattern, predict when the user is most likely to experience withdrawal symptoms, and prevent a relapse by offering tailored, personalized intervention. The ability of the platform to deliver contextually appropriate text messages and seamlessly integrate Just-in-Time treatment delivery into smokers’ daily lives will help make the transition from a smoker to an ex-smoker, easy and effective.
Our state-of-the-art algorithms detect smoking gestures using sensors in your smartwatch.
Delivering the right intervention at the right time begins with the ability of the platform to automatically detect when a person in smoking. We achieve this by sensing hand to mouth gestures using a wristband sensor. Our initial pilot study was able to detect smoking gestures in real-time with an accuracy of 95.7%. Since data from the wristband sensors are continuously analyzed, we can automatically and unobtrusively determine when the user is smoking. The video above shows the reconstruction of an arm movement of a person who is smoking while walking, from data collected using a wristband sensor.
When a person smokes, often without realizing, they connect smoking with things. These then become triggers; environmental factors or contextual cues that act as precipitants of a smoking event. Some common triggers are people, locations, and time of the day. Automatically identifying these are key to prevent smoking lapses. Using data from a smartphone, sophisticated machine learning algorithm, and data analytics, our platform starts predicting these triggers for each user after a short learning period.
Our technology helps us predict and identify a high-risk situation when the user is likely to lapse and fall back to smoking. We prevent these lapses by delivering personalized intervention in the form of a motivational text message that provides personalized strategies and tips to handle cravings . The technology and the intervention for Just-in-Time treatment delivery platform is designed by a team of distinguished doctors and engineers. The system is being validated by running a clinical trial with 100 subjects.
This platform has room to tailor the intervention protocol to the needs of the user. Along with personalized motivational messages, a high-risk situation could also trigger a phone call from a therapist, remind the user to take medication, change a patch, or trigger a call from a friend. Our ability to automate and personalize therapy creates numerous options for the user to create a quit program that works for them.
The long-term goal of Lumme is to develop reliable devices that can detect and treat a wide range of behavioral disorders. The first issue we are trying to solve in this multidimensional problem is smoking. Going forward, the capability of recognizing other activities such as eating and drinking using hand-to-mouth gesture recognition will be explored. This would not only enable us to study the relation between these behaviors and smoking, but also study other disorders. The extension of this technology can be employed to detect and treat eating disorders such as obesity, binge-eating, and anorexia. At Lumme, we aspire to develop a comprehensive platform to revolutionize the detection and treatment of addictive behaviors and disorders.