Electrocardiogram (ECG) signal evaluation is routinely used in clinics as a significant diagnostic method for detecting arrhythmia. However, it is very labor intensive to externally evaluate ECG signals, due to their small amplitude. Using automated detection and classification methods in the clinic can assist doctors in making accurate and expeditious diagnoses of diseases. In this study, we developed a classification method for arrhythmia based on the combination of a convolutional neural network and long short-term memory, which was then used to diagnose eight ECG signals, including a normal sinus rhythm. The ECG data of the experiment were derived from the MIT-BIH arrhythmia database. The experimental method mainly consisted of two parts. The input data of the model were two-dimensional grayscale images converted from one-dimensional signals, and detection and classification of the input data was carried out using the combined model. The advantage of this method is that it does not require performing feature extraction or noise filtering on the ECG signal. The experimental results showed that the implemented method demonstrated high classification performance in terms of accuracy, specificity, and sensitivity equal to 99.01%, 99.57%, and 97.67%, respectively. Our proposed model can assist doctors in accurately detecting arrhythmia during routine ECG screening.

Atrial fibrillation (AF) is the most common persistent arrhythmia and is likely to cause strokes and damage to heart function in patients. Electrocardiogram (ECG) is the gold standard for detecting AF. However, ECGs have short boards with short monitoring cycles and problems with gathering. It is also difficult to detect a burst AF through ECG. In contrast, photoplethysmography (PPG) is easy to perform and suitable for long-term monitoring. In this study, we propose a method that combines time-frequency analysis with deep learning and identifies AF based on PPG. The advantage of the method is that there is no need for the noise filtering and feature extraction of PPG, and it has a high generalization capability. The data for the experiment came from three publicly accessible databases. The first part of the experimental method uses data augmentation to convert the 10 s PPG segment into a time-frequency chromatograph by means of time-frequency analysis. The second part inputs the chromatograph into a hybrid framework that combines a convolutional neural network (CNN) and long short-term memory (LSTM) for AF/nonAF classification. The experimental results show that the method has a high classification accuracy, sensitivity, specificity, and F1 score, which are equal to 98.21%, 98.00%, 98.07% and 98.13%, respectively. The area under the receiver operating characteristic curve (AUC) is 0.9959. The model we propose not only aids doctors in diagnosing AF but also provides a method for identifying AF through portable wearable devices.

Sleep apnea (SA) is a common sleep disorder which could impair the human physiological system. Therefore, early diagnosis of SA is of great interest. The traditional method of diagnosing SA is an overnight polysomnography (PSG) evaluation. When PSG has limited availability, automatic SA screening with a fewer number of signals should be considered. The primary purpose of this study is to develop and evaluate a SA detection model based on electrocardiogram (ECG) and blood oxygen saturation (SpO2). We adopted a multimodal approach to fuse ECG and SpO2 signals at the feature level. Then, feature selection was conducted using the recursive feature elimination with cross-validation (RFECV) algorithm and random forest (RF) classifier used to discriminate between apnea and normal events. Experiments were conducted on the Apnea-ECG database. The introduced algorithm obtained an accuracy of 97.5%, a sensitivity of 95.9%, a specificity of 98.4% and an AUC of 0.992 in per-segment classification, and outperformed previous works. The results showed that ECG and SpO2 are complementary in detecting SA, and that the combination of ECG and SpO2 enhances the ability to diagnose SA. Therefore, the proposed method has the potential to be an alternative to conventional detection methods.

The measurement of hemoglobin is a vital index for diagnosing and monitoring diseases in clinical practice. At present, solutions need to be found for the soreness, high risk of infection, and inconvenient operation associated with invasive detection methods. This paper proposes a method for non-invasively detecting hemoglobin levels based on multi-wavelength photoplethysmography (PPG) signals. AFE4490 and TMUX1109 were used to implement the low-cost collection of an eight-LED transmissive PPG signal. We used seven regular LEDs and one broadband LED (Osram SFH4737) as light sources. Additionally, a finger clip integrating multiple sensors was designed and manufactured via 3D printing to simultaneously monitor the LED–sensor distance and the pressure from the tester’s finger during PPG signal acquisition. We used a method to extract features from PPG signals using a sliding-window’s variance and an evaluation metric for PPG signals based on the AdaCost classification. Data were gathered from 56 participants from the Nephrology department, including 16 anemic patients. Pearson correlation analysis was conducted on the collected data to remove any data with a weak correlation. The advantage of using a broadband LED as a light source was also demonstrated. Several non-invasive hemoglobin regression models were created by applying AdaBoost, BPNN, and Random Forest models. The study’s results indicate that the AdaBoost model produced the best performance, with a mean absolute error (MAE) of 2.67 g/L and a correlation coefficient (R²) of 0.91 The study results show that the device we designed and manufactured can achieve effective non-invasive hemoglobin detection and represents a new methodological approach to obtaining measurements that can be applied in a clinical setting.