Recently, smartphones with mobile health applications have become promising tools in the healthcare industry due to their convenience, ubiquity for patients, and the ability to gather data in real time. In this paper, we propose a novel non-invasive, portable, and cuff-less method for monitoring BP by only using the smartphones’ camera. Our experiment uses pulse transit time (PTT) between two separate photoplethysmogram (PPG) signals to estimate the subjects’ systolic blood pressure (SBP) and diastolic blood pressure (DBP). Our proposed method first measures the subject’s PPG signals from his/her index fingers using the smartphones’ camera. Then, filtering and peak detection algorithms of the proposed method reduce the motion and noise artifacts in the PPG signals. Finally, the proposed method estimates SBP and DBP based on a linear regression model which was trained and tested on 30 trials with six healthy subjects. We evaluated the proposed method by comparing BP values of the proposed method with those of the reference (or gold-standard) device in terms of mean absolute error (MAE), standard deviation of error (SD), and R-squared (R2) value of the cross-validation. Experimental results show that the proposed method estimates the average of MAE ± SD is 2.07 ± 2.06 mm Hg for SBP estimation, and 2.12 ± 1.85 mm Hg for DBP estimation. These estimates are lower than accurate BP estimation standard (5 ± 8 mmHg).

Bicycle riders are exposed to accident injuries such as head trauma. The risk of these riders’ injuries is higher compared to the risk of injuries for motorists. Crashes, riders’ errors, and environmental hazards are the cause of bicycle-related accidents. In 2017, nearly 50% of bicycle-related accidents occurred in urban areas at night, which may contribute to a delay in reporting the accidents to emergency centers. Hence, a system that can detect the accident is needed to notify urgent care clinics promptly. In this paper, we propose a bicycle accident detection system. We designed hardware modules measuring the features related to the riding status of a bicycle and fall accidents. For this purpose, we used a magnetic, angular rate, and gravity (MARG) sensor-based system which measures four different types of signals: 1) acceleration, 2) angular velocity, 3) angle, and 4) magnitude of the riding status. Each of these signals is measured in three different directions (X,Y, and Z). We used two different time-domain parameters, i.e., average and standard deviation. As a result, we considered 24 features. We used principal component analysis (PCA) for feature reduction and the support vector machines (SVM) algorithm for the detection of fall accidents. Experimental results show that our proposed system detects fall accidents during cycling status with 95.2% accuracy, which demonstrates the feasibility of our proposed bicycle accident detection system.

Photoplethysmography (PPG) is a technique to detect blood volume changes in an optical way. Representative PPG applications are the measurements of oxygen saturation, heart rate, and respiratory rate. However, the PPG signals are sensitive to motion and noise artifacts (MNAs), especially when they are obtained from smartphone cameras. Moreover, the PPG signals are different among users and each individual's PPG signal has a unique characteristic. Hence, an effective MNA detection and reduction method for smartphone PPG signals, which adapts itself to each user in a personalized way, is highly demanded. In this paper, a concept of the probabilistic neural network is introduced to be used with the proposed extracted parameters. The signal amplitude, standard deviation of peak to peak time intervals and amplitudes, along with the mean of moving standard deviation, signal slope changes, and the optimal autoregressive model order are proposed for effective MNA detection. Accordingly, the performance of the proposed personalized algorithm is compared with conventional MNA detection algorithms. As for the performance metrics, we considered accuracy, sensitivity, and specificity. The results show that the overall performance of the personalized MNA detection is enhanced compared to the generalized algorithm. The average values of the accuracy, sensitivity, and specificity of the personalized one are 98.07%, 92.6%, and 99.78%, respectively, while these are 89.92%, 84.21%, and 93.63% for the general one.

We have recently found that our previously-developed atrial fibrillation (AF) detection algorithm for smartphones can give false positives when subjects' fingers or hands move, as we rely on proper finger placement over the smartphone camera to collect the signal of interest. Specifically, smartphone camera pulsatile signals that are obtained from normal sinus rhythm (NSR) subjects but are corrupted by motion and noise artifacts (MNAs) are frequently detected as AF. AF and motion-corrupted episodes have the similar characteristic that pulse-to-pulse intervals are irregular. We have developed an MNA-resilient smartphone-based AF detection algorithm that first discriminates and eliminates MNA-corrupted episodes in smartphone camera recordings, and then detects AF in MNA-free recordings. We found that MNA-corrupted episodes have highly varying pulse slope, large turning point ratio, or large kurtosis values in smartphone signals compared to MNA-free AF and NSR episodes. We first use these three metrics for MNA discrimination and exclusion. Then, AF is detected in MNA-free signals using our previous algorithm. The capability to discriminate MNAs and AFs separately in smartphone signals increases the specificity of AF detection. To evaluate the performance of the proposed MNA-resilient AF algorithm, 99 subjects, including 88 study participants with AF at baseline and in NSR after electrical cardioversion as well as 11 participants with MNA-corrupted NSR, were recruited. Using iPhone 4S, 5S, and 6S models, we collected 2-min pulsatile time series from each subject. The clinical results show that the accuracy, sensitivity, and specificity of the proposed AF algorithm are 0.97, 0.98, and 0.97, respectively, which are higher than those of the previous AF algorithm.

Keratoconus (KC) is a progressive eye disease and a major cause of vision impairment and blindness worldwide. Early diagnosis is crucial for effective management, yet conventional diagnostic methods rely on expensive and bulky imaging devices, limiting accessibility, especially in resource-constrained settings. This paper introduces a novel smartphone-based approach for the early detection of KC, leveraging screen-projected Placido disc patterns and an advanced image processing framework. Unlike traditional corneal topographers, our method utilizes a unique Placido disc projection technique and a machine learning-based classification model to analyze corneal irregularities with high precision. With a sensitivity of 96.08%, specificity of 97.96%, and overall accuracy of 97% on our dataset, the proposed system demonstrates exceptional diagnostic reliability. By transforming a standard smartphone into an effective screening tool, this innovation provides an affordable, portable, and user-friendly solution for early KC detection, bridging the gap in eye care accessibility and reducing the global burden of undiagnosed keratoconus.