Refining Time-Activity Classification of Human Subjects Using the Global Positioning System

Automated classification of time-activity-location patterns for improved estimation of personal exposure to air pollution

BACKGROUND

Air pollution epidemiology has primarily relied on measurements from fixed outdoor air quality monitoring stations to derive population-scale exposure. Characterisation of individual time-activity-location patterns is critical for accurate estimations of personal exposure and dose because pollutant concentrations and inhalation rates vary significantly by location and activity.

METHODS

We developed and evaluated an automated model to classify major exposure-related microenvironments (home, work, other static, in-transit) and separated them into indoor and outdoor locations, sleeping activity and five modes of transport (walking, cycling, car, bus, metro/train) with multidisciplinary methods from the fields of movement ecology and artificial intelligence. As input parameters, we used GPS coordinates, accelerometry, and noise, collected at 1 min intervals with a validated Personal Air quality Monitor (PAM) carried by 35 volunteers for one week each. The model classifications were then evaluated against manual time-activity logs kept by participants.

RESULTS

Overall, the model performed reliably in classifying home, work, and other indoor microenvironments (F1-score>0.70) but only moderately well for sleeping and visits to outdoor microenvironments (F1-score=0.57 and 0.3 respectively). Random forest approaches performed very well in classifying modes of transport (F1-score>0.91). We found that the performance of the automated methods significantly surpassed those of manual logs.

CONCLUSIONS

Automated models for time-activity classification can markedly improve exposure metrics. Such models can be developed in many programming languages, and if well formulated can have general applicability in large-scale health studies, providing a comprehensive picture of environmental health risks during daily life with readily gathered parameters from smartphone technologies.

Particulate air pollution in the Copenhagen metro part 2: Low-cost sensors and micro-environment classification

In this study fine particulate matter (PM_2.5) levels throughout the Copenhagen metro system are measured for the first time and found to be ∼10 times the roadside levels in Copenhagen. In this Part 2 article, low-cost sensor (LCS) nodes designed for personal-exposure monitoring are tested against a conventional mid-range device (TSI DustTrak), and gravimetric methods. The nodes were found to be effective for personal exposure measurements inside the metro system, with R² values of > 0.8 at 1-min and > 0.9 at 5-min time-resolution, with an average slope of 1.01 in both cases, in comparison to the reference, which is impressive for this dynamic environment. Micro-environment (ME) classification techniques are also developed and tested, involving the use of auxiliary sensors, measuring light, carbon dioxide, humidity, temperature and motion. The output from these sensors is used to distinguish between specific MEs, namely, being aboard trains travelling above- or under- ground, with 83 % accuracy, and determining whether sensors were aboard a train or stationary at a platform with 92 % accuracy. This information was used to show a 143 % increase in mean PM_2.5 concentration for underground sections relative to overground, and 22 % increase for train vs. platform measurements. The ME classification method can also be used to improve calibration models, assist in accurate exposure assessment based on detailed time-activity patterns, and facilitate field studies that do not require personnel to record time-activity diaries.

Application of machine learning to predict transport modes from GPS, accelerometer, and heart rate data

BACKGROUND

There has been an increased focus on active transport, but the measurement of active transport is still difficult and error-prone. Sensor data have been used to predict active transport. While heart rate data have very rarely been considered before, this study used random forests (RF) to predict transport modes using Global Positioning System (GPS), accelerometer, and heart rate data and paid attention to methodological issues related to the prediction strategy and post-processing.

METHODS

The RECORD MultiSensor study collected GPS, accelerometer, and heart rate data over seven days from 126 participants living in the Ile-de-France region. RF models were built to predict transport modes for every minute (ground truth information on modes is from a GPS-based mobility survey), splitting observations between a Training dataset and a Test dataset at the participant level instead at the minute level. Moreover, several window sizes were tested for the post-processing moving average of the predicted transport mode.

RESULTS

The minute-level prediction rate of being on trips vs. at a visited location was 90%. Final prediction rates of transport modes ranged from 65% for public transport to 95% for biking. Using minute-level observations from the same participants in the Training and Test sets (as RF spontaneously does) upwardly biases prediction rates. The inclusion of heart rate data improved prediction rates only for biking. A 3 to 5-min bandwidth moving average was optimum for a posteriori homogenization.

CONCLUSION

Heart rate only very slightly contributed to better predictions for specific transport modes. Moreover, our study shows that Training and Test sets must be carefully defined in RF models and that post-processing with carefully chosen moving average windows can improve predictions.

Dynamic classification of personal microenvironments using a suite of wearable, low-cost sensors

Human exposure to air pollution is associated with increased risk of morbidity and mortality. However, personal air pollution exposures can vary substantially depending on an individual's daily activity patterns and air quality within their residence and workplace. This work developed and validated an adaptive buffer size (ABS) algorithm capable of dynamically classifying an individual's time spent in predefined microenvironments using data from global positioning systems (GPS), motion sensors, temperature sensors, and light sensors. Twenty-two participants in Fort Collins, CO were recruited to carry a personal air sampler for a 48-h period. The personal sampler was retrofitted with a GPS and a pushbutton to complement the existing sensor measurements (temperature, motion, light). The pushbutton was used in conjunction with a traditional time-activity diary to note when the participant was located at "home", "work", or within an "other" microenvironment. The ABS algorithm predicted the amount of time spent in each microenvironment with a median accuracy of 99.1%, 98.9%, and 97.5% for the "home", "work", and "other" microenvironments. The ability to classify microenvironments dynamically in real time can enable the development of new sampling and measurement technologies that classify personal exposure by microenvironment.

The Use of Inertial Measurement Units for the Study of Free Living Environment Activity Assessment: A Literature Review

This article presents an overview of fifty-eight articles dedicated to the evaluation of physical activity in free-living conditions using wearable motion sensors. This review provides a comprehensive summary of the technical aspects linked to sensors (types, number, body positions, and technical characteristics) as well as a deep discussion on the protocols implemented in free-living conditions (environment, duration, instructions, activities, and annotation). Finally, it presents a description and a comparison of the main algorithms and processing tools used for assessing physical activity from raw signals.

Towards Personal Exposures: How Technology Is Changing Air Pollution and Health Research

PURPOSE OF REVIEW

We present a review of emerging technologies and how these can transform personal air pollution exposure assessment and subsequent health research.

RECENT FINDINGS

Estimating personal air pollution exposures is currently split broadly into methods for modeling exposures for large populations versus measuring exposures for small populations. Air pollution sensors, smartphones, and air pollution models capitalizing on big/new data sources offer tremendous opportunity for unifying these approaches and improving long-term personal exposure prediction at scales needed for population-based research. A multi-disciplinary approach is needed to combine these technologies to not only estimate personal exposures for epidemiological research but also determine drivers of these exposures and new prevention opportunities. While available technologies can revolutionize air pollution exposure research, ethical, privacy, logistical, and data science challenges must be met before widespread implementations occur. Available technologies and related advances in data science can improve long-term personal air pollution exposure estimates at scales needed for population-based research. This will advance our ability to evaluate the impacts of air pollution on human health and develop effective prevention strategies.

A systematic review of data mining and machine learning for air pollution epidemiology

Background

Data measuring airborne pollutants, public health and environmental factors are increasingly being stored and merged. These big datasets offer great potential, but also challenge traditional epidemiological methods. This has motivated the exploration of alternative methods to make predictions, find patterns and extract information. To this end, data mining and machine learning algorithms are increasingly being applied to air pollution epidemiology.

Methods

We conducted a systematic literature review on the application of data mining and machine learning methods in air pollution epidemiology. We carried out our search process in PubMed, the MEDLINE database and Google Scholar. Research articles applying data mining and machine learning methods to air pollution epidemiology were queried and reviewed.

Results

Our search queries resulted in 400 research articles. Our fine-grained analysis employed our inclusion/exclusion criteria to reduce the results to 47 articles, which we separate into three primary areas of interest: 1) source apportionment; 2) forecasting/prediction of air pollution/quality or exposure; and 3) generating hypotheses. Early applications had a preference for artificial neural networks. In more recent work, decision trees, support vector machines, k-means clustering and the APRIORI algorithm have been widely applied. Our survey shows that the majority of the research has been conducted in Europe, China and the USA, and that data mining is becoming an increasingly common tool in environmental health. For potential new directions, we have identified that deep learning and geo-spacial pattern mining are two burgeoning areas of data mining that have good potential for future applications in air pollution epidemiology.

Conclusions

We carried out a systematic review identifying the current trends, challenges and new directions to explore in the application of data mining methods to air pollution epidemiology. This work shows that data mining is increasingly being applied in air pollution epidemiology.

The potential to support air pollution epidemiology continues to grow with advancements in data mining related to temporal and geo-spacial mining, and deep learning. This is further supported by new sensors and storage mediums that enable larger, better quality data. This suggests that many more fruitful applications can be expected in the future.

Integrating data from multiple time-location measurement methods for use in exposure assessment: the Multi-Ethnic Study of Atherosclerosis and Air Pollution (MESA Air)

Tools to assess time-location patterns related to environmental exposures have expanded from reliance on time-location diaries (TLDs) and questionnaires to use of geospatial location devices such as data-logging Global Positioning System (GPS) equipment. The Multi-Ethnic Study of Atherosclerosis and Air Pollution obtained typical time-location patterns via questionnaire for 6424 adults in six US cities. At a later time (mean 4.6 years after questionnaire), a subset (n=128) participated in high-resolution data collection for specific 2-week periods resulting in concurrent GPS and detailed TLD data, which were aggregated to estimate time spent in various microenvironments. During these 2-week periods, participants were observed to spend the most time at home indoors (mean of 78%) and a small proportion of time in-vehicle (mean of 4%). Similar overall patterns were reported by these participants on the prior questionnaire (mean home indoors: 75%; mean in-vehicle: 4%). However, individual micro-environmental time estimates measured over specific 2-week periods were not highly correlated with an individual's questionnaire report of typical behavior (Spearman's ρ of 0.43 for home indoors and 0.39 for in-vehicle). Although questionnaire data about typical time-location patterns can inform interpretation of long-term epidemiological analyses and risk assessment, they may not reliably represent an individual's short-term experience.

Ethical implications of location and accelerometer measurement in health research studies with mobile sensing devices

Quantification of individual behaviours using mobile sensing devices, including physical activity and spatial location, is a rapidly growing field in both academic research and the corporate world. In this case study, we summarize the literature examining the ethical aspects of mobile sensing and argue that a robust discussion about the ethical implications of mobile sensing for research purposes has not occurred sufficiently in the literature. Based on our literature summary and guided by basic ethical principles set out in Canadian, US, and International Ethics documents we propose four areas where further discussion should occur: consent, privacy and confidentiality, mitigating risk, and consideration of vulnerable populations. We argue that ongoing consent is crucial for participants to be aware of the precision and volume of data that is collected with mobile sensing devices. Related to privacy we discuss that participants may not agree that anonymized data is sufficient for privacy and confidentiality when mobile sensing data are collected. There has been some discussion about mitigating risk in the literature. We highlight that the researchers' obligations toward mitigating risks that are not directly related to the study purpose are unclear and require considerable discussion. Finally, using mobile sensing devices to study vulnerable populations requires careful consideration, particularly with respect to balancing research needs with participant burden. Based on our discussion, we identify a broad set of unanswered questions about the ethics of mobile sensing that should be addressed by the research community.

How Sensors Might Help Define the External Exposome

The advent of the exposome concept, the advancement of mobile technology, sensors, and the “internet of things” bring exciting opportunities to exposure science. Smartphone apps, wireless devices, the downsizing of monitoring technologies, along with lower costs for such equipment makes it possible for various aspects of exposure to be measured more easily and frequently. We discuss possibilities and lay out several criteria for using smart technologies for external exposome studies. Smart technologies are evolving quickly, and while they provide great promise for advancing exposure science, many are still in developmental stages and their use in epidemiology and risk studies must be carefully considered. The most useable technologies for exposure studies at this time relate to gathering exposure-factor data, such as location and activities. Development of some environmental sensors (e.g., for some air pollutants, noise, UV) is moving towards making the use of these more reliable and accessible to research studies. The possibility of accessing such an unprecedented amount of personal data also comes with various limitations and challenges, which are discussed. The advantage of improving the collection of long term exposure factor data is that this can be combined with more “traditional” measurement data to model exposures to numerous environmental factors.