A Novel Personalized Motion and Noise Artifact (MNA) Detection Method for Smartphone Photoplethysmograph (PPG) Signals

Accuracy enhancement of metabolic index-based blood glucose estimation with a screening process for low-quality data

Significance

Many researchers have proposed various non-invasive glucose monitoring (NIGM) approaches using wearable or portable devices. However, due to the limited capacity of detectors for such compact devices and the movement of the body during measurement, the precision of the acquired data frequently diminishes, which can cause problems during actual use in daily life. In addition, intensive smoothing is often used in post-processing to mitigate the effects of erroneous values. However, this requires a considerable amount of data and results in a delay in the response to the actual blood glucose level (BGL).

Aim

Instead of just applying data smoothing in the post-process of the data acquisition, we propose an active low-quality data screening method in the pre-process. In the proposal phase of the screening process, we employ an analytical approach to examine and formulate factors that might affect the BGL estimation accuracy.

Approach

A signal quality index inspired by the standard deviation concept is introduced to detect visually apparent noise on signals. Furthermore, the total estimation error in the metabolic index (MI) is calculated based on potential perturbations defined by the signal-to-noise ratio (SNR) and the uncertainty due to discrete sampling. Thereafter, the acquired data were screened by these quality indices.

Results

By applying the proposed data screening process to the data obtained from a commercially available smartwatch device in the pre-process, the estimation accuracy of the MI-based BGL was improved significantly.

Conclusions

Adopting the proposed screen process improves BGL estimation accuracy in the smartwatch-based prototype. Applying the proposed screen process will facilitate the integration of wearable and continuous BGL monitoring into size- and SNR-limited devices such as smartwatches and smart rings.

A Novel Machine-Learning-Based Noise Detection Method for Photoplethysmography Signals

Wearable devices are widespread for continuous health monitoring; capturing various physiological parameters for remote health monitoring and early detection of health issues. These devices are susceptible to interference such as Motion Artifacts (MA) and Baseline Wanders (BW). Mitigating potential false alarms due to those artifacts is an important challenge in wearable healthcare. To tackle this challenge, it is crucial to first identify noise in the signals recorded by wearable systems. Most of the conventional methods rely on reference data like accelerometer data to detect noise in Photoplethysmogram (PPG) signals. This study proposes a Machine Learning (ML)-based approach to distinguish between clean and corrupted segments in PPG signals without relying on other sensors' data. Binary and three-class classification on clean, MA-, and BW-corrupted signals produce promising F1-scores from 89.3% to 99.4%.

A calibration method for smartphone camera photophlethysmography

Smartphone camera photoplethysmography (cPPG) enables non-invasive pulse oximetry and hemoglobin concentration measurements. However, the aesthetic-driven non-linearity in default image capture and preprocessing pipelines poses challenges for consistency and transferability of cPPG across devices. This work identifies two key parameters-tone mapping and sensor threshold-that significantly impact cPPG measurements. We propose a novel calibration method to linearize camera measurements, thus enhancing consistency and transferability of cPPG across devices. A benchtop calibration system is also presented, leveraging a microcontroller and LED setup to characterize these parameters for each phone model. Our validation studies demonstrate that, with appropriate calibration and camera settings, cPPG applications can achieve 74% higher accuracy than with default settings. Moreover, our calibration method proves effective across different smartphone models (N = 4 ), and calibrations performed on one phone can be applied to other smartphones of the same model (N = 6 ), enhancing consistency and scalability of cPPG applications.

Artifacts classification and apnea events detection in neck photoplethysmography signals

The novel pulse oximetry measurement site of the neck is a promising location for multi-modal physiological monitoring. Specifically, in the context of respiratory monitoring, in which it is important to have direct information about airflow. The neck makes this possible, in contrast to common photoplethysmography (PPG) sensing sites. However, this PPG signal is susceptible to artifacts that critically impair the signal quality. To fully exploit neck PPG for reliable physiological parameters extraction and apneas monitoring, this paper aims to develop two classification algorithms for artifacts and apnea detection. Features from the time, correlogram, and frequency domains were extracted. Two SVM classifiers with RBF kernels were trained for different window (W) lengths and thresholds (Thd) of corruption. For artifacts classification, the maximum performance was attained for the parameters combination of [W = 6s-Thd= 20%], with an average accuracy= 85.84%(ACC), sensitivity= 85.43%(SE) and specificity= 86.26%(SP). For apnea detection, the model [W = 10s-Thd= 50%] maximized all the performance metrics significantly (ACC= 88.25%, SE= 89.03%, SP= 87.42%). The findings of this proof of concept are significant for denoising novel neck PPG signals, and demonstrate, for the first time, that it is possible to promptly detect apnea events from neck PPG signals in an instantaneous manner. This could make a big impact in crucial real-time applications, like devices to prevent sudden-unexpected-death-in-epilepsy (SUDEP).

User Authentication Recognition Process Using Long Short-Term Memory Model

User authentication (UA) is the process by which biometric techniques are used by a person to gain access to a physical or virtual site. UA has been implemented in various applications such as financial transactions, data privacy, and access control. Various techniques, such as facial and fingerprint recognition, have been proposed for healthcare monitoring to address biometric recognition problems. Photoplethysmography (PPG) technology is an optical sensing technique which collects volumetric blood change data from the subject’s skin near the fingertips, earlobes, or forehead. PPG signals can be readily acquired from devices such as smartphones, smartwatches, or web cameras. Classical machine learning techniques, such as decision trees, support vector machine (SVM), and k-nearest neighbor (kNN), have been proposed for PPG identification. We developed a UA classification method for smart devices using long short-term memory (LSTM). Specifically, our UA classifier algorithm uses raw signals so as not to lose the specific characteristics of the PPG signal coming from each user’s specific behavior. In the UA context, false positive and false negative rates are crucial. We recruited thirty healthy subjects and used a smartphone to take PPG data. Experimental results show that our Bi-LSTM-based UA algorithm based on the feature-based machine learning and raw data-based deep learning approaches provides 95.0% and 96.7% accuracy, respectively.

A supervised machine learning semantic segmentation approach for detecting artifacts in plethysmography signals from wearables

Objective. Wearable devices equipped with plethysmography (PPG) sensors provided a low-cost, long-term solution to early diagnosis and continuous screening of heart conditions. However PPG signals collected from such devices often suffer from corruption caused by artifacts. The objective of this study is to develop an effective supervised algorithm to locate the regions of artifacts within PPG signals.Approach. We treat artifact detection as a 1D segmentation problem. We solve it via a novel combination of an active-contour-based loss and an adapted U-Net architecture. The proposed algorithm was trained on the PPG DaLiA training set, and further evaluated on the PPG DaLiA testing set, WESAD dataset and TROIKA dataset.Main results. We evaluated with the DICE score, a well-established metric for segmentation accuracy evaluation in the field of computer vision. The proposed method outperforms baseline methods on all three datasets by a large margin (≈7 percentage points above the next best method). On the PPG DaLiA testing set, WESAD dataset and TROIKA dataset, the proposed method achieved 0.8734 ± 0.0018, 0.9114 ± 0.0033 and 0.8050 ± 0.0116 respectively. The next best method only achieved 0.8068 ± 0.0014, 0.8446 ± 0.0013 and 0.7247 ± 0.0050.Significance. The proposed method is able to pinpoint exact locations of artifacts with high precision; in the past, we had only a binary classification of whether a PPG signal has good or poor quality. This more nuanced information will be critical to further inform the design of algorithms to detect cardiac arrhythmia.

Breathing rate estimation based on multiple linear regression

The breathing rate is a key clinical parameter that can now be estimated using photoplethysmographic methods. Here, we present an indirect method of breathing rate estimation that does not require bulky and uncomfortable sensors. Breathing modulates a pulsed wave; we extracted the maximum and minimum values, and first-order derivatives thereof, to measure breathing amplitude, frequency, and baseline drift. Demodulation was used to obtain multiple breathing waveforms, from which peak values are extracted to obtain breathing rates. Multiple linear regression was used to combine the breathing rates of different feature points. We used a breathing dataset for 53 subjects, and divided the data into training and test sets when calculating the regression coefficients. We also assessed the generalizability of our linear model. We found that breathing rate estimation was more accurate when using a multivariate signal method with multiple versus a single feature point. The mean absolute error, mean error, and standard deviation of the error were 1.28, -0.07, and 1.60 breaths per minute, respectively.

Non-Contact HR Monitoring via Smartphone and Webcam During Different Respiratory Maneuvers and Body Movements

As a reliable indicator for individual's healthiness conditions, heart rate (HR) has been widely considered and used. Imaging photoplethysmography (iPPG) is recently highlighted as a promising HR measurement method, due to its non-contact characteristics, by extracting the HR from facial video recordings. In this study, we propose a camera-based HR monitoring technique that estimates HR information from iPPG signals extracted from a video sequence. Videos were recorded using a smartphone or a laptop camera. We adopted the plane-orthogonal-to-skin (POS) method to compute iPPG. The proposed method is evaluated by applying it to extract HR of 9 subjects at rest and during two motion conditions (lateral and frontal) while they were performing several respiratory maneuvers-spontaneous, metronome, and forced. Automatic face detection algorithms were implemented in the proposed method. Our experimental results show that mean values of HR have 0.56% error and 99.4% accuracy when compared to HR calculated from the gold-standard electrocardiography (ECG) reference in diverse conditions of motions and respiratory maneuvers.

Optical difference in the frequency domain to suppress disturbance for wearable electronics

Measurements based on optics offer a wide range of unprecedented opportunities in the biological application due to the noninvasive or non-destructive detection. Wearable skin-like optoelectronic devices, capable of deforming with the human skin, play significant roles in future biomedical engineering such as clinical diagnostics or daily healthcare. However, the detected signals based on light intensity are very sensitive to the light path. The performance degradation of the wearable devices occurs due to device deformation or motion artifact. In this work, we propose the optical difference in the frequency domain of signals for suppressing the disturbance generated by wearable device deformation or motion artifact during the photoplethysmogram (PPG) monitoring. The signal processing is simulated with different input waveforms for analyzing the performance of this method. Then we design and fabricate a wearable optoelectronic device to monitor the PPG signal in the condition of motion artifact and use the optical difference in the frequency domain of signals to suppress irregular disturbance. The proposed method reduced the average error in heart rate estimation from 13.04 beats per minute (bpm) to 3.41 bpm in motion and deformation situations. These consequences open up a new prospect for improving the performance of the wearable optoelectronic devices and precise medical monitoring in the future.

Novel Image Processing Method for Detecting Strep Throat (Streptococcal Pharyngitis) Using Smartphone

In this paper, we propose a novel strep throat detection method using a smartphone with an add-on gadget. Our smartphone-based strep throat detection method is based on the use of camera and flashlight embedded in a smartphone. The proposed algorithm acquires throat image using a smartphone with a gadget, processes the acquired images using color transformation and color correction algorithms, and finally classifies streptococcal pharyngitis (or strep) throat from healthy throat using machine learning techniques. Our developed gadget was designed to minimize the reflection of light entering the camera sensor. The scope of this paper is confined to binary classification between strep and healthy throats. Specifically, we adopted k-fold validation technique for classification, which finds the best decision boundary from training and validation sets and applies the acquired best decision boundary to the test sets. Experimental results show that our proposed detection method detects strep throats with 93.75% accuracy, 88% specificity, and 87.5% sensitivity on average.

Monitoring of Heart Rate from Photoplethysmographic Signals Using a Samsung Galaxy Note8 in Underwater Environments

Photoplethysmography (PPG) is a commonly used in determining heart rate and oxygen saturation (SpO₂). However, PPG measurements and its accuracy are heavily affected by the measurement procedure and environmental factors such as light, temperature, and medium. In this paper, we analyzed the effects of different mediums (water vs. air) and temperature on the PPG signal quality and heart rate estimation. To evaluate the accuracy, we compared our measurement output with a gold-standard PPG device (NeXus-10 MKII). The experimental results show that the average PPG signal amplitude values of the underwater environment decreased considerably (22% decrease) compared to PPG signals of dry environments, and the heart rate measurement deviated 7% (5 beats per minute on average. The experimental results also show that the signal to noise ratio (SNR) and signal amplitude decrease as temperature decreases. Paired t-test which compares amplitude and heart rate values between the underwater and dry environments was performed and the test results show statistically significant differences for both amplitude and heart rate values (p < 0.05). Moreover, experimental results indicate that decreasing the temperature from 45 °C to 5 °C or changing the medium from air to water decreases PPG signal quality, (e.g., PPG signal amplitude decreases from 0.560 to 0.112). The heart rate is estimated within 5.06 bpm deviation at 18 °C in underwater environment, while estimation accuracy decreases as temperature goes down.

A novel diversity method for smartphone camera-based heart rhythm signals in the presence of motion and noise artifacts

The advent of smartphones has advanced the use of embedded sensors to acquire various physiological information. For example, smartphone camera sensors and accelerometers can provide heart rhythm signals to the subjects, while microphones can give respiratory signals. However, the acquired smartphone-based physiological signals are more vulnerable to motion and noise artifacts (MNAs) compared to using medical devices, since subjects need to hold the smartphone with proper contact to the smartphone camera and lens stably and tightly for a duration of time without any movement in the hand or finger. This results in more MNA than traditional methods, such as placing a finger inside a tightly enclosed pulse oximeter to get PPG signals, which provides stable contact between the sensor and the subject's finger. Moreover, a smartphone lens does not block ambient light in an effective way, while pulse oximeters are designed to block the ambient light effectively. In this paper, we propose a novel diversity method for smartphone signals that reduces the effect of MNAs during heart rhythm signal detection by 1) acquiring two heterogeneous signals from a color intensity signal and a fingertip movement signal, and 2) selecting the less MNA-corrupted signal of the two signals. The proposed method has advantages in that 1) diversity gain can be obtained from the two heterogeneous signals when one signal is clean while the other signal is corrupted, and 2) acquisition of the two heterogeneous signals does not double the acquisition procedure but maintains a single acquisition procedure, since two heterogeneous signals can be obtained from a single smartphone camera recording. In our diversity method, we propose to choose the better signal based on the signal quality indices (SQIs), i.e., standard deviation of instantaneous heart rate (STD-HR), root mean square of the successive differences of peak-to-peak time intervals (RMSSD-T), and standard deviation of peak values (STD-PV). As a performance metric evaluating the proposed diversity method, the ratio of usable period is considered. Experimental results show that our diversity method increases the usable period 19.53% and 6.25% compared to the color intensity or the fingertip movement signals only, respectively.