Wearable Heart Rate monitors are used in sports to provide physiological insights into athletes’ well-being and performance. Their unobtrusive nature and ability to provide reliable heart rate measurements facilitate the estimation of cardiorespiratory fitness of athletes, as quantified by maximum consumption of oxygen uptake. Previous studies have employed data-driven models which use heart rate information to estimate the cardiorespiratory fitness of athletes. This signifies the physiological relevance of heart rate and heart rate variability for the estimation of maximal oxygen uptake. In this work, the heart rate variability features that were extracted from both exercise and recovery segments were fed to three different Machine Learning models to estimate maximal oxygen uptake of 856 athletes performing Graded Exercise Testing. A total of 101 features from exercise and 30 features from recovery segments were given as input to three feature selection methods to avoid overfitting of the models and to obtain relevant features. This resulted in the increase of model’s accuracy by 5.7% for exercise and 4.3% for recovery. Further, post-modelling analysis was performed to remove the deviant points in two cases, initially in both training and testing and then only in training set, using k-Nearest Neighbour. In the former case, the removal of deviant points led to a reduction of 19.3% and 18.0% in overall estimation error for exercise and recovery, respectively. In the latter case, which mimicked the real-world scenario, the average R value of the models was observed to be 0.72 and 0.70 for exercise and recovery, respectively. From the above experimental approach, the utility of heart rate variability to estimate maximal oxygen uptake of large population of athletes was validated. Additionally, the proposed work contributes to the utility of cardiorespiratory fitness assessment of athletes through wearable heart rate monitors.

Respiration rate is a vital parameter to indicate good health, wellbeing, and performance. As the estimation through classical measurement modes are limited only to rest or during slow movements, respiration rate is commonly estimated through physiological signals such as electrocardiogram and photoplethysmography due to the unobtrusive nature of wearable devices. Deep learning methodologies have gained much traction in the recent past to enhance accuracy during activities involving a lot of movement. However, these methods pose challenges, including model interpretability, uncertainty estimation in the context of respiration rate estimation, and model compactness in terms of deployment in wearable platforms. In this direction, we propose a multifunctional framework, which includes the combination of an attention mechanism, an uncertainty estimation functionality, and a knowledge distillation framework. We evaluated the performance of our framework on two datasets containing ambulatory movement. The attention mechanism visually and quantitatively improved instantaneous respiration rate estimation. Using Monte Carlo dropouts to embed the network with inferential uncertainty estimation resulted in the rejection of 3.7% of windows with high uncertainty, which consequently resulted in an overall reduction of 7.99% in the mean absolute error. The attention-aware knowledge distillation mechanism reduced the model’s parameter count and inference time by 49.5% and 38.09%, respectively, without any increase in error rates. Through experimentation, ablation, and visualization, we demonstrated the efficacy of the proposed framework in addressing practical challenges, thus taking a step towards deployment in wearable edge devices.

Deep learning networks are being developed in every stage of the MRI workflow and have provided state-of-the-art results. However, this has come at the cost of increased computation requirement and storage. Hence, replacing the networks with compact models at various stages in the MRI workflow can significantly reduce the required storage space and provide considerable speedup. In computer vision, knowledge distillation is a commonly used method for model compression. In our work, we propose a knowledge distillation (KD) framework for the image to image problems in the MRI workflow in order to develop compact, low-parameter models without a significant drop in performance. We propose a combination of the attention-based feature distillation method and imitation loss and demonstrate its effectiveness on the popular MRI reconstruction architecture, DC-CNN. We conduct extensive experiments using Cardiac, Brain, and Knee MRI datasets for 4x, 5x and 8x accelerations. We observed that the student network trained with the assistance of the teacher using our proposed KD framework provided significant improvement over the student network trained without assistance across all the datasets and acceleration factors. Specifically, for the Knee dataset, the student network achieves $65\%$ parameter reduction, 2x faster CPU running time, and 1.5x faster GPU running time compared to the teacher. Furthermore, we compare our attention-based feature distillation method with other feature distillation methods. We also conduct an ablative study to understand the significance of attention-based distillation and imitation loss. We also extend our KD framework for MRI super-resolution and show encouraging results.

Cardiac arrhythmia is a prevalent and significant cause of morbidity and mortality among cardiac ailments. Early diagnosis is crucial in providing intervention for patients suffering from cardiac arrhythmia. Traditionally, diagnosis is performed by examination of the Electrocardiogram (ECG) by a cardiologist. This method of diagnosis is hampered by the lack of accessibility to expert cardiologists. For quite some time, signal processing methods had been used to automate arrhythmia diagnosis. However, these traditional methods require expert knowledge and are unable to model a wide range of arrhythmia. Recently, Deep Learning methods have provided solutions to performing arrhythmia diagnosis at scale. However, the black-box nature of these models prohibit clinical interpretation of cardiac arrhythmia. There is a dire need to correlate the obtained model outputs to the corresponding segments of the ECG. To this end, two methods are proposed to provide interpretability to the models. The first method is a novel application of Gradient-weighted Class Activation Map (Grad-CAM) for visualizing the saliency of the CNN model. In the second approach, saliency is derived by learning the input deletion mask for the LSTM model. The visualizations are provided on a model whose competence is established by comparisons against baselines. The results of model saliency not only provide insight into the prediction capability of the model but also aligns with the medical literature for the classification of cardiac arrhythmia.

Continuous monitoring of blood oxygen saturation levels is vital for patients with pulmonary disorders. Traditionally, SpO$_2$ monitoring has been carried out using transmittance pulse oximeters due to its dependability. However, SpO$_2$ measurement from transmittance pulse oximeters is limited to peripheral regions. This becomes a disadvantage at very low temperatures as blood perfusion to the peripherals decreases. On the other hand, reflectance pulse oximeters can be used at various sites like finger, wrist, chest and forehead. Additionally, reflectance pulse oximeters can be scaled down to affordable patches that do not interfere with the user's diurnal activities. However, accurate SpO$_2$ estimation from reflectance pulse oximeters is challenging due to its patient dependent, subjective nature of measurement. Recently, a Machine Learning (ML) method was used to model reflectance waveforms onto SpO$_2$ obtained from transmittance waveforms. However, the generalizability of the model to new patients was not tested. In light of this, the current work implemented multiple ML based approaches which were subsequently found to be incapable of generalizing to new patients. Furthermore, a minimally calibrated data driven approach was utilized in order to obtain SpO$_2$ from reflectance PPG waveforms. The proposed solution produces an average mean absolute error of 1.81\% on unseen patients which is well within the clinically permissible error of 2\%. Two statistical tests were conducted to establish the effectiveness of the proposed method.

The exponential rise in wearable sensors has garnered significant interest in assessing the physiological parameters during day-to-day activities. Respiration rate is one of the vital parameters used in the performance assessment of lifestyle activities. However, obtrusive setup for measurement, motion artifacts, and other noises complicate the process. This paper presents a multitasking architecture based on Deep Learning (DL) for estimating instantaneous and average respiration rate from ECG and accelerometer signals, such that it performs efficiently under daily living activities like cycling, walking, etc. The multitasking network consists of a combination of Encoder-Decoder and Encoder-IncResNet, to fetch the average respiration rate and the respiration signal. The respiration signal can be leveraged to obtain the breathing peaks and instantaneous breathing cycles. Mean absolute error(MAE), Root mean square error (RMSE), inference time, and parameter count analysis has been used to compare the network with the current state of art Machine Learning (ML) model and other DL models developed in previous studies. Other DL configurations based on a variety of inputs are also developed as a part of the work. The proposed model showed better overall accuracy and gave better results than individual modalities during different activities.