Abstract. Advances in ambient environmental monitoring technologies are enabling concerned communities and citizens to collect data to better understand their local environment and potential exposures. These mobile, low-cost tools make it possible to collect data with increased temporal and spatial resolution, providing data on a large scale with unprecedented levels of detail. This type of data has the potential to empower people to make personal decisions about their exposure and support the development of local strategies for reducing pollution and improving health outcomes. However, calibration of these low-cost instruments has been a challenge. Often, a sensor package is calibrated via field calibration. This involves colocating the sensor package with a high-quality reference instrument for an extended period and then applying machine learning or other model fitting technique such as multiple linear regression to develop a calibration model for converting raw sensor signals to pollutant concentrations. Although this method helps to correct for the effects of ambient conditions (e.g., temperature) and cross sensitivities with nontarget pollutants, there is a growing body of evidence that calibration models can overfit to a given location or set of environmental conditions on account of the incidental correlation between pollutant levels and environmental conditions, including diurnal cycles. As a result, a sensor package trained at a field site may provide less reliable data when moved, or transferred, to a different location. This is a potential concern for applications seeking to perform monitoring away from regulatory monitoring sites, such as personal mobile monitoring or high-resolution monitoring of a neighborhood. We performed experiments confirming that transferability is indeed a problem and show that it can be improved by collecting data from multiple regulatory sites and building a calibration model that leverages data from a more diverse data set. We deployed three sensor packages to each of three sites with reference monitors (nine packages total) and then rotated the sensor packages through the sites over time. Two sites were in San Diego, CA, with a third outside of Bakersfield, CA, offering varying environmental conditions, general air quality composition, and pollutant concentrations. When compared to prior single-site calibration, the multisite approach exhibits better model transferability for a range of modeling approaches. Our experiments also reveal that random forest is especially prone to overfitting and confirm prior results that transfer is a significant source of both bias and standard error. Linear regression, on the other hand, although it exhibits relatively high error, does not degrade much in transfer. Bias dominated in our experiments, suggesting that transferability might be easily increased by detecting and correcting for bias. Also, given that many monitoring applications involve the deployment of many sensor packages based on the same sensing technology, there is an opportunity to leverage the availability of multiple sensors at multiple sites during calibration to lower the cost of training and better tolerate transfer. We contribute a new neural network architecture model termed split-NN that splits the model into two stages, in which the first stage corrects for sensor-to-sensor variation and the second stage uses the combined data of all the sensors to build a model for a single sensor package. The split-NN modeling approach outperforms multiple linear regression, traditional two- and four-layer neural networks, and random forest models. Depending on the training configuration, compared to random forest the split-NN method reduced error 0 %–11 % for NO2 and 6 %–13 % for O3.
more »
« less
Accuracy and Validity of a Single Inertial Measurement Unit-Based System to Determine Upper Limb Kinematics for Medically Underserved Populations
Stroke is one of the leading causes of death and disability worldwide, with a disproportionate burden represented by low- and middle-income countries (LMICs). To improve post-stroke outcomes in LMICs, researchers have sought to leverage emerging technologies that overcome traditional barriers associated with stroke management. One such technology, inertial measurement units (IMUs), exhibit great potential as a low-cost, portable means to evaluate and monitor patient progress during decentralized rehabilitation protocols. As such, the aim of the present study was to determine the ability of a low-cost single IMU sensor-based wearable system (named the T’ena sensor) to reliably and accurately assess movement quality and efficiency in physically and neurologically healthy adults. Upper limb movement kinematics measured by the T’ena sensor were compared to the gold standard reference system during three functional tasks, and root mean square errors, Pearson’s correlation coefficients, intraclass correlation coefficients, and the Bland Altman method were used to compare kinematic variables of interest between the two systems for absolute accuracy and equivalency. The T’ena sensor and the gold standard reference system were significantly correlated for all tasks and measures ( r range = 0.648—0.947), although less so for the Finger to Nose task ( r range = 0.648—0.894). Results demonstrate that single IMU systems are a valid, reliable, and objective method by which to measure movement kinematics during functional tasks. Context-appropriate enabling technologies specifically designed to address barriers to quality health services in LMICs can accelerate progress towards the United Nations Sustainable Development Goal 3.
more »
« less
- Award ID(s):
- 1752255
- NSF-PAR ID:
- 10404252
- Date Published:
- Journal Name:
- Frontiers in Bioengineering and Biotechnology
- Volume:
- 10
- ISSN:
- 2296-4185
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Background: Over 10% of antibiotics in low- and middle-income countries (LMICs) are substandard or falsified. Detection of poor-quality antibiotics via the gold standard method, high-performance liquid chromatography (HPLC), is slow and costly. Paper analytical devices (PADs) and antibiotic paper analytical devices (aPADs) have been developed as an inexpensive way to estimate antibiotic quality in LMICs. Aim: To model the impact of using a rapid screening tools, PADs/aPADs, to improve the quality of amoxicillin used for treatment of childhood pneumonia in Kenya. Methods: We developed an agent-based model, ESTEEM (Examining Screening Technologies with Economic Evaluations for Medicines), to estimate the effectiveness and cost savings of incorporating PADs and aPADs in amoxicillin quality surveillance in Kenya. We compared the current testing scenario (batches of entire samples tested by HPLC) with an expedited HPLC scenario (testing smaller batches at a time), as well as a screening scenario using PADs/aPADs to identify poor-quality amoxicillin followed by confirmatory analysis with HPLC. Results: Scenarios using PADs/aPADs or expedited HPLC yielded greater incremental benefits than the current testing scenario by annually averting 586 (90% uncertainty range (UR) 364–874) and 221 (90% UR 126–332) child pneumonia deaths, respectively. The PADs/aPADs screening scenario identified and removed poor-quality antibiotics faster than the expedited or regular HPLC scenarios, and reduced costs significantly. The PADs/aPADs scenario resulted in an incremental return of $14.9 million annually compared with the reference scenario of only using HPLC. Conclusion: This analysis shows the significant value of PADs/aPADs as a medicine quality screening and testing tool in LMICs with limited resources.more » « less
-
Parkinson’s disease (PD) is a neurological progressive movement disorder, affecting more than 10 million people globally. PD demands a longitudinal assessment of symptoms to monitor the disease progression and manage the treatments. Existing assessment methods require patients with PD (PwPD) to visit a clinic every 3–6 months to perform movement assessments conducted by trained clinicians. However, periodic visits pose barriers as PwPDs have limited mobility, and healthcare cost increases. Hence, there is a strong demand for using telemedicine technologies for assessing PwPDs in remote settings. In this work, we present an in-home telemedicine kit, named iTex (intelligent Textile), which is a patient-centered design to carry out accessible tele-assessments of movement symptoms in people with PD. iTex is composed of a pair of smart textile gloves connected to a customized embedded tablet. iTex gloves are integrated with flex sensors on the fingers and inertial measurement unit (IMU) and have an onboard microcontroller unit with IoT (Internet of Things) capabilities including data storage and wireless communication. The gloves acquire the sensor data wirelessly to monitor various hand movements such as finger tapping, hand opening and closing, and other movement tasks. The gloves are connected to a customized tablet computer acting as an IoT device, configured to host a wireless access point, and host an MQTT broker and a time-series database server. The tablet also employs a patient-centered interface to guide PwPDs through the movement exam protocol. The system was deployed in four PwPDs who used iTex at home independently for a week. They performed the test independently before and after medication intake. Later, we performed data analysis of the in-home study and created a feature set. The study findings reported that the iTex gloves were capable to collect movement-related data and distinguish between pre-medication and post-medication cases in a majority of the participants. The IoT infrastructure demonstrated robust performance in home settings and offered minimum barriers for the assessment exams and the data communication with a remote server. In the post-study survey, all four participants expressed that the system was easy to use and poses a minimum barrier to performing the test independently. The present findings indicate that the iTex glove system has the potential for periodic and objective assessment of PD motor symptoms in remote settings.more » « less
-
Functional connectivity between the brain and body kinematics has largely not been investigated due to the requirement of motionlessness in neuroimaging techniques such as functional magnetic resonance imaging (fMRI). However, this connectivity is disrupted in many neurodegenerative disorders, including Parkinson’s Disease (PD), a neurological progressive disorder characterized by movement symptoms including slowness of movement, stiffness, tremors at rest, and walking and standing instability. In this study, brain activity is recorded through functional near-infrared spectroscopy (fNIRS) and electroencephalography (EEG), and body kinematics were captured by a motion capture system (Mocap) based on an inertial measurement unit (IMU) for gross movements (large movements such as limb kinematics), and the WearUp glove for fine movements (small range movements such as finger kinematics). PD and neurotypical (NT) participants were recruited to perform 8 different movement tasks. The recorded data from each modality have been analyzed individually, and the processed data has been used for classification between the PD and NT groups. The average changes in oxygenated hemoglobin (HbO2) from fNIRS, EEG power spectral density in the Theta, Alpha, and Beta bands, acceleration vector from Mocap, and normalized WearUp flex sensor data were used for classification. 12 different support vector machine (SVM) classifiers have been used on different datasets such as only fNIRS data, only EEG data, hybrid fNIRS/EEG data, and all the fused data for two classification scenarios: classifying PD and NT based on individual activities, and all activity data fused together. The PD and NT group could be distinguished with more than 83% accuracy for each individual activity. For all the fused data, the PD and NT groups are classified with 81.23%, 92.79%, 92.27%, and 93.40% accuracy for the fNIRS only, EEG only, hybrid fNIRS/EEG, and all fused data, respectively. The results indicate that the overall performance of classification in distinguishing PD and NT groups improves when using both brain and body data.more » « less
-
null (Ed.)The objective of this study was to test the feasibility of using a pair of wearable inertial measurement unit (IMU) sensors to accurately capture dynamic joint motion data during simulated occupational conditions. Eleven subjects (5 males and 6 females) performed repetitive neck, low-back, and shoulder motions simulating low- and high-difficulty occupational tasks in a laboratory setting. Kinematics for each of the 3 joints were measured via IMU sensors in addition to a “gold standard” passivemarker optical motion capture system. The IMU accuracy was benchmarked relative to the optical motion capture system, and IMU sensitivity to low- and high-difficulty tasks was evaluated. The accuracy of the IMU sensors was found to be very good on average, but significant positional drift was observed in some trials. In addition, IMU measurements were shown to be sensitive to differences in task difficulty in all 3 joints (P < .05). These results demonstrate the feasibility for using wearable IMU sensors to capture kinematic exposures as potential indicators of occupational injury risk. Velocities and accelerations demonstrate the most potential for developing risk metrics since they are sensitive to task difficulty and less sensitive to drift than rotational position measurements.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.3389/fbioe.2022.918617
