Towards a generalizable method for detecting fluid intake with wrist-mounted sensors and adaptive segmentation

Abstract

Over the last decade, advances in mobile technologies have enabled the development of intelligent systems that attempt to recognize and model a variety of health-related human behaviors. While automated dietary monitoring based on passive sensors has been an area of increasing research activity for many years, much less attention has been given to tracking fluid intake. In this work, we apply an adaptive segmentation technique on a continuous stream of inertial data captured with a practical, off-the-shelf wrist-mounted device to detect fluid intake gestures passively. We evaluated our approach in a study with 30 participants where 561 drinking instances were recorded. Using a leave-one-participant-out (LOPO), we were able to detect drinking episodes with 90.3% precision and 91.0% recall, demonstrating the generalizability of our approach. In addition to our proposed method, we also contribute an anonymized and labeled dataset of drinking and non-drinking gestures to encourage further work in the field.

Fluid Intake Monitoring Systems for the Elderly: A Review of the Literature

“…Several wearable techniques are proposed to detect food or liquid intake including inertial measurements of the wrist or body [ 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 , 54 , 55 , 56 , 57 , 58 , 59 , 60 , 61 , 62 , 63 , 64 , 65 , 66 , 67 , 68 , 69 , 70 , 71 , 72 , 73 ], or textile-based measurements [ 74 , 75 , 76 , 77 , 78 , 79 , 80 , 81 , 82 , 83 , 84 , 85 , 86 ]. The majority of studies analyzed and classified food intake with less focus on liquid intake.…”

“…Chun et al focused on detecting drinking using adaptive segmentation with a single commercial wrist sensor with 30 participants drinking from four containers and performing a multitude of ADLs including eating and drinking [ 65 ]. They achieved an average precision 90.3% and recall 91% to detect drinking with binary classification [ 65 ].…”

“…Chun et al focused on detecting drinking using adaptive segmentation with a single commercial wrist sensor with 30 participants drinking from four containers and performing a multitude of ADLs including eating and drinking [ 65 ]. They achieved an average precision 90.3% and recall 91% to detect drinking with binary classification [ 65 ]. Instead of using the traditional fixed length sliding window, they used an adaptive approach which allowed them match the exact duration of the drinking instances [ 65 ].…”

“…They achieved an average precision 90.3% and recall 91% to detect drinking with binary classification [ 65 ]. Instead of using the traditional fixed length sliding window, they used an adaptive approach which allowed them match the exact duration of the drinking instances [ 65 ]. They found the RF model performed the best with LOSO cross validation, using 45 extracted features [ 65 ].…”

Fluid Intake Monitoring System Using a Wearable Inertial Sensor for Fluid Intake Management

“…The best performance of classification was 86% accuracy using random forest (RF). Chun et al [ 26 ] employed an adaptive segmentation technique and various machine-learning-based classifiers to detect fluid intake. The system obtained the best results of 90.3% precision and 91.0% sensitivity by using the RF model.…”