Researchers from Northwestern University have developed a smart necklace which can help one quit smoking, and is becoming imminent. It is a smart neck-worn device resembling a lapis blue pendant that detects a user's smoking much more reliably than previous systems, by capturing heat signatures from thermal sensors. Lapis blue is greenish blue to violetish blue, and a  highly saturated colour. 


The study describing the necklace was published February 13 in the Proceedings of the ACM on Interactive, Mobile, Wearable, and Ubiquitous Technologies. 


How does the smart necklace work?


The pendant tracks heat signatures from lit cigarettes in real time, and how much smokers inhale, and the time between puffs. The goal of the smart necklace is to intervene to prevent smokers from relapsing after they quit. 


The necklace is called SmokeMon. It completely maintains a smoker's privacy, and only tracks heat, and not the visuals. This is a critical factor for people to feel comfortable wearing it. 


The authors noted in the paper that SmokeMon is a chest-worn thermal sensing wearable system that can capture spatial, temporal and thermal information around the wearer and cigarette all day to unobtrusively and passively detect smoking events. 


What is smoking topography?


Details such as when a person lights a cigarette, when they hold it to their mouth and take a puff, how much they inhale, how much time they take between two consecutive puffs, and how long they have the cigarette in their mouth are called smoking topography. 


Smoking topography is important for two reasons. Firstly, it allows scientists to measure and assess harmful carbon monoxide exposure among smokers and understand more deeply the relationship between chemical exposure and tobacco-related diseases including cancer, heart disease, lung disease, stroke, chronic obstructive pulmonary disease (COPD), diabetes, chronic bronchitis and emphysema. 


The second reason why smoking topography is important is that it helps people in their efforts to quit smoking by understanding how smoking topography relates to relapse. This happens frequently in people who quit. 


Information on how many puffs a former smoker takes can be used to intervene with a phone call to encourage the person to prevent a relapse. The researchers are planning to study the effectiveness of the necklace in detecting smoking puffs from electronic cigarettes.


Cigarette smoke contains thousands of chemicals that are harmful and cause tobacco-related diseases, and to date, the causality between human exposure to specific compounds and the harmful effects is unknown.  


Each year, smoking kills more than eight million people worldwide, and remains a leading cause of preventable disease, disability and death in the United States, where smoking accounts for more than 4,80,000 deaths annually. 


How is SmokeMon better than existing devices?


Existing devices to track smoking topography are attached to the cigarette. This can change how a person smokes, making the data less reliable. Some researchers have tried to measure smoking behaviour using non-obtrusive ways such as the use of wrist-worn inertial measurement unit sensors in smartwatches. However, these approaches can generate several false positives because they are often confounded by non-smoking hand-to-mouth gestures. 


While motion-based wearable sensors and their corresponding machine-learned models have shown promise in unobtrusively tracking smoking gestures, they confound smoking with other similar hand-to-mouth gestures such as eating and drinking. 


Wearable video cameras are another option. However, they create privacy and stigma concerns. This limits the applicability of camera-based approaches in natural settings. 


The authors noted that the current gold-standard approaches to smoking topography involve expensive, bulky and obtrusive sensor devices, which create unnatural smoking behaviour and prevent the potential for real-time interventions in the wild.


How was the study conducted?


As many as 19 participants took part in the research. They participated in 115 smoking sessions in which scientists examined their smoking behaviour in controlled and free-living experiments. 


The scientists trained a deep-learning based machine model that could detect smoking events along with their smoking topography, while the participants wore the device. The things the necklace could detect include the timing of the puff, number of puffs, puff duration, puff volume, inter-puff interval, and smoking duration. In order to understand how the participants felt about the device, the researchers also ran three focus groups with 18 tobacco-treatment specialists.  


The researchers evaluated SmokeKon in both controlled and free-living experiments with a total of 19 participants, more than 110 hours of data, and 115 smoking sessions. 


SmokeMon demonstrated a good performance


The device achieved an F1-score (a machine learning evaluation metric that measures a model's accuracy) of 0.9 for puff detection in the laboratory, and 0.8 in the wild. 


The authors conducted that by providing SmokeMon as an open platform, they provide measurement of smoking topography in free-living settings to enable testing of smoking topography in the real world. Therefore, the authors noted, SmokeMon has the potential to facilitate timely smoking cessation interventions.