UAdam: Unified Adam-Type Algorithmic Framework for Nonconvex Optimization.

Adam-type algorithms have become a preferred choice for optimization in the deep learning setting; however, despite their success, their convergence is still not well understood. To this end, we introduce a unified framework for Adam-type algorithms, termed UAdam. It is equipped with a general form of the second-order moment, which makes it possible to include Adam and its existing and future variants as special cases, such as NAdam, AMSGrad, AdaBound, AdaFom, and Adan. The approach is supported by a rigorous convergence analysis of UAdam in the general nonconvex stochastic setting, showing that UAdam converges to the neighborhood of stationary points with a rate of O(1/T). Furthermore, the size of the neighborhood decreases as the parameter ß1 increases. Importantly, our analysis only requires the first-order momentum factor to be close enough to 1, without any restrictions on the second-order momentum factor. Theoretical results also reveal the convergence conditions of vanilla Adam, together with the selection of appropriate hyperparameters. This provides a theoretical guarantee for the analysis, applications, and further developments of the whole general class of Adam-type algorithms. Finally, several numerical experiments are provided to support our theoretical findings.

Graph-Regularized Tensor Regression: A Domain-Aware Framework for Interpretable Modeling of Multiway Data on Graphs.

Modern data analytics applications are increasingly characterized by exceedingly large and multidimensional data sources. This represents a challenge for traditional machine learning models, as the number of model parameters needed to process such data grows exponentially with the data dimensions, an effect known as the curse of dimensionality. Recently, tensor decomposition (TD) techniques have shown promising results in reducing the computational costs associated with large-dimensional models while achieving comparable performance. However, such tensor models are often unable to incorporate the underlying domain knowledge when compressing high-dimensional models. To this end, we introduce a novel graph-regularized tensor regression (GRTR) framework, whereby domain knowledge about intramodal relations is incorporated into the model in the form of a graph Laplacian matrix. This is then used as a regularization tool to promote a physically meaningful structure within the model parameters. By virtue of tensor algebra, the proposed framework is shown to be fully interpretable, both coefficient-wise and dimension-wise. The GRTR model is validated in a multiway regression setting and compared against competing models and is shown to achieve improved performance at reduced computational costs. Detailed visualizations are provided to help readers gain an intuitive understanding of the employed tensor operations.

An In-Ear PPG-Based Blood Glucose Monitor: A Proof-of-Concept Study.

Monitoring diabetes saves lives. To this end, we introduce a novel, unobtrusive, and readily deployable in-ear device for the continuous and non-invasive measurement of blood glucose levels (BGLs). The device is equipped with a low-cost commercially available pulse oximeter whose infrared wavelength (880 nm) is used for the acquisition of photoplethysmography (PPG). For rigor, we considered a full range of diabetic conditions (non-diabetic, pre-diabetic, type I diabetic, and type II diabetic). Recordings spanned nine different days, starting in the morning while fasting, up to a minimum of a two-hour period after eating a carbohydrate-rich breakfast. The BGLs from PPG were estimated using a suite of regression-based machine learning models, which were trained on characteristic features of PPG cycles pertaining to high and low BGLs. The analysis shows that, as desired, an average of 82% of the BGLs estimated from PPG lie in region A of the Clarke error grid (CEG) plot, with 100% of the estimated BGLs in the clinically acceptable CEG regions A and B. These results demonstrate the potential of the ear canal as a site for non-invasive blood glucose monitoring.

Design and Implementation of an Atrial Fibrillation Detection Algorithm on the ARM Cortex-M4 Microcontroller.

At present, a medium-level microcontroller is capable of performing edge computing and can handle the computation of neural network kernel functions. This makes it possible to implement a complete end-to-end solution incorporating signal acquisition, digital signal processing, and machine learning for the classification of cardiac arrhythmias on a small wearable device. In this work, we describe the design and implementation of several classifiers for atrial fibrillation detection on a general-purpose ARM Cortex-M4 microcontroller. We used the CMSIS-DSP library, which supports Naïve Bayes and Support Vector Machine classifiers, with different kernel functions. We also developed Python scripts to automatically transfer the Python model (trained in Scikit-learn) to the C environment. To train and evaluate the models, we used part of the data from the PhysioNet/Computing in Cardiology Challenge 2020 and performed simple classification of atrial fibrillation based on heart-rate irregularity. The performance of the classifiers was tested on a general-purpose ARM Cortex-M4 microcontroller (STM32WB55RG). Our study reveals that among the tested classifiers, the SVM classifier with RBF kernel function achieves the highest accuracy of 96.9%, sensitivity of 98.4%, and specificity of 95.8%. The execution time of this classifier was 720 µs per recording. We also discuss the advantages of moving computing tasks to edge devices, including increased power efficiency of the system, improved patient data privacy and security, and reduced overall system operation costs. In addition, we highlight a problem with false-positive detection and unclear significance of device-detected atrial fibrillation.

A Pilot Study of Heart Rate Variability Synchrony as a Marker of Intraoperative Surgical Teamwork and Its Correlation to the Length of Procedure.

Objective: Quality of intraoperative teamwork may have a direct impact on patient outcomes. Heart rate variability (HRV) synchrony may be useful for objective assessment of team cohesion and good teamwork. The primary aim of this study was to investigate the feasibility of using HRV synchrony in surgical teams. Secondary aims were to investigate the association of HRV synchrony with length of procedure (LOP), complications, number of intraoperative glitches and length of stay (LOS). We also investigated the correlation between HRV synchrony and team familiarity, pre- and intraoperative stress levels (STAI questionnaire), NOTECHS score and experience of team members. Methods: Ear, nose and throat (ENT) and vascular surgeons (consultant and registrar team members) were recruited into the study. Baseline demographics including level of team members' experience were gathered before each procedure. For each procedure, continuous electrocardiogram (ECG) recording was performed and questionnaires regarding pre- and intraoperative stress levels and non-technical skills (NOTECHS) scores were collected for each team member. An independent observer documented the time of each intraoperative glitch. Statistical analysis was conducted using stepwise multiple linear regression. Results: Four HRV synchrony metrics which may be markers of efficient surgical collaboration were identified from the data: 1. number of HRV synchronies per hour of procedure, 2. number of HRV synchrony trends per hour of procedure, 3. length of HRV synchrony trends per hour of procedure, 4. area under the HRV synchrony trend curve per hour of procedure. LOP was inversely correlated with number of HRV synchrony trends per hour of procedure (p < 0.0001), area under HRV synchrony trend curve per hour of procedure (p = 0.001), length of HRV synchrony trends per hour of procedure (p = 0.002) and number of HRV synchronies per hour of procedure (p < 0.0001). LOP was positively correlated with: FS (p = 0.043; R = 0.358) and intraoperative STAI score of the whole team (p = 0.007; R = 0.493). Conclusions: HRV synchrony metrics within operating teams may be used as an objective marker to quantify surgical teamwork. We have shown that LOP is shorter when the intraoperative surgical teams' HRV is more synchronised.

Multivariate Multiscale Cosine Similarity Entropy and Its Application to Examine Circularity Properties in Division Algebras.

The extension of sample entropy methodologies to multivariate signals has received considerable attention, with traditional univariate entropy methods, such as sample entropy (SampEn) and fuzzy entropy (FuzzyEn), introduced to measure the complexity of chaotic systems in terms of irregularity and randomness. The corresponding multivariate methods, multivariate multiscale sample entropy (MMSE) and multivariate multiscale fuzzy entropy (MMFE), were developed to explore the structural richness within signals at high scales. However, the requirement of high scale limits the selection of embedding dimension and thus, the performance is unavoidably restricted by the trade-off between the data size and the required high scale. More importantly, the scale of interest in different situations is varying, yet little is known about the optimal setting of the scale range in MMSE and MMFE. To this end, we extend the univariate cosine similarity entropy (CSE) method to the multivariate case, and show that the resulting multivariate multiscale cosine similarity entropy (MMCSE) is capable of quantifying structural complexity through the degree of self-correlation within signals. The proposed approach relaxes the prohibitive constraints between the embedding dimension and data length, and aims to quantify the structural complexity based on the degree of self-correlation at low scales. The proposed MMCSE is applied to the examination of the complex and quaternion circularity properties of signals with varying correlation behaviors, and simulations show the MMCSE outperforming the standard methods, MMSE and MMFE.

Variational Embedding Multiscale Sample Entropy: A Tool for Complexity Analysis of Multichannel Systems.

Entropy-based methods have received considerable attention in the quantification of structural complexity of real-world systems. Among numerous empirical entropy algorithms, conditional entropy-based methods such as sample entropy, which are associated with amplitude distance calculation, are quite intuitive to interpret but require excessive data lengths for meaningful evaluation at large scales. To address this issue, we propose the variational embedding multiscale sample entropy (veMSE) method and conclusively demonstrate its ability to operate robustly, even with several times shorter data than the existing conditional entropy-based methods. The analysis reveals that veMSE also exhibits other desirable properties, such as the robustness to the variation in embedding dimension and noise resilience. For rigor, unlike the existing multivariate methods, the proposed veMSE assigns a different embedding dimension to every data channel, which makes its operation independent of channel permutation. The veMSE is tested on both stimulated and real world signals, and its performance is evaluated against the existing multivariate multiscale sample entropy methods. The proposed veMSE is also shown to exhibit computational advantages over the existing amplitude distance-based entropy methods.

In-Ear SpO2: A Tool for Wearable, Unobtrusive Monitoring of Core Blood Oxygen Saturation.

The non-invasive estimation of blood oxygen saturation (SpO2) by pulse oximetry is of vital importance clinically, from the detection of sleep apnea to the recent ambulatory monitoring of hypoxemia in the delayed post-infective phase of COVID-19. In this proof of concept study, we set out to establish the feasibility of SpO2 measurement from the ear canal as a convenient site for long term monitoring, and perform a comprehensive comparison with the right index finger-the conventional clinical measurement site. During resting blood oxygen saturation estimation, we found a root mean square difference of 1.47% between the two measurement sites, with a mean difference of 0.23% higher SpO2 in the right ear canal. Using breath holds, we observe the known phenomena of time delay between central circulation and peripheral circulation with a mean delay between the ear and finger of 12.4 s across all subjects. Furthermore, we document the lower photoplethysmogram amplitude from the ear canal and suggest ways to mitigate this issue. In conjunction with the well-known robustness to temperature induced vasoconstriction, this makes conclusive evidence for in-ear SpO2 monitoring being both convenient and superior to conventional finger measurement for continuous non-intrusive monitoring in both clinical and everyday-life settings.

A Novel Multivariate Sample Entropy Algorithm for Modeling Time Series Synchronization.

Approximate and sample entropy (AE and SE) provide robust measures of the deterministic or stochastic content of a time series (regularity), as well as the degree of structural richness (complexity), through operations at multiple data scales. Despite the success of the univariate algorithms, multivariate sample entropy (mSE) algorithms are still in their infancy and have considerable shortcomings. Not only are existing mSE algorithms unable to analyse within- and cross-channel dynamics, they can counter-intuitively interpret increased correlation between variates as decreased regularity. To this end, we first revisit the embedding of multivariate delay vectors (DVs), critical to ensuring physically meaningful and accurate analysis. We next propose a novel mSE algorithm and demonstrate its improved performance over existing work, for synthetic data and for classifying wake and sleep states from real-world physiological data. It is furthermore revealed that, unlike other tools, such as the correlation of phase synchrony, synchronized regularity dynamics are uniquely identified via mSE analysis. In addition, a model for the operation of this novel algorithm in the presence of white Gaussian noise is presented, which, in contrast to the existing algorithms, reveals for the first time that increasing correlation between different variates reduces entropy.

Physiological artifacts in scalp EEG and ear-EEG.

BACKGROUND: A problem inherent to recording EEG is the interference arising from noise and artifacts. While in a laboratory environment, artifacts and interference can, to a large extent, be avoided or controlled, in real-life scenarios this is a challenge. Ear-EEG is a concept where EEG is acquired from electrodes in the ear. METHODS: We present a characterization of physiological artifacts generated in a controlled environment for nine subjects. The influence of the artifacts was quantified in terms of the signal-to-noise ratio (SNR) deterioration of the auditory steady-state response. Alpha band modulation was also studied in an open/closed eyes paradigm. RESULTS: Artifacts related to jaw muscle contractions were present all over the scalp and in the ear, with the highest SNR deteriorations in the gamma band. The SNR deterioration for jaw artifacts were in general higher in the ear compared to the scalp. Whereas eye-blinking did not influence the SNR in the ear, it was significant for all groups of scalps electrodes in the delta and theta bands. Eye movements resulted in statistical significant SNR deterioration in both frontal, temporal and ear electrodes. Recordings of alpha band modulation showed increased power and coherence of the EEG for ear and scalp electrodes in the closed-eyes periods. CONCLUSIONS: Ear-EEG is a method developed for unobtrusive and discreet recording over long periods of time and in real-life environments. This study investigated the influence of the most important types of physiological artifacts, and demonstrated that spontaneous activity, in terms of alpha band oscillations, could be recorded from the ear-EEG platform. In its present form ear-EEG was more prone to jaw related artifacts and less prone to eye-blinking artifacts compared to state-of-the-art scalp based systems.

Multi-scale pixel-based image fusion using multivariate empirical mode decomposition.

A novel scheme to perform the fusion of multiple images using the multivariate empirical mode decomposition (MEMD) algorithm is proposed. Standard multi-scale fusion techniques make a priori assumptions regarding input data, whereas standard univariate empirical mode decomposition (EMD)-based fusion techniques suffer from inherent mode mixing and mode misalignment issues, characterized respectively by either a single intrinsic mode function (IMF) containing multiple scales or the same indexed IMFs corresponding to multiple input images carrying different frequency information. We show that MEMD overcomes these problems by being fully data adaptive and by aligning common frequency scales from multiple channels, thus enabling their comparison at a pixel level and subsequent fusion at multiple data scales. We then demonstrate the potential of the proposed scheme on a large dataset of real-world multi-exposure and multi-focus images and compare the results against those obtained from standard fusion algorithms, including the principal component analysis (PCA), discrete wavelet transform (DWT) and non-subsampled contourlet transform (NCT). A variety of image fusion quality measures are employed for the objective evaluation of the proposed method. We also report the results of a hypothesis testing approach on our large image dataset to identify statistically-significant performance differences.

The Deep-Match Framework: R-Peak Detection in Ear-ECG.

The Ear-ECG provides a continuous Lead I like electrocardiogram (ECG) by measuring the potential difference related to heart activity by electrodes which are embedded within earphones. However, the significant increase in wearability and comfort enabled by Ear-ECG is often accompanied by a degradation in signal quality - an obstacle that is shared by the majority of wearable technologies. We aim to resolve this issue by introducing a Deep Matched Filter (Deep-MF) for the highly accurate detection of R-peaks in wearable ECG, thus enhancing the utility of Ear-ECG in real-world scenarios. The Deep-MF consists of an encoder stage, partially initialised with an ECG template, and an R-peak classifier stage. Through its operation as a Matched Filter, the encoder searches for matches with an ECG template in the input signal, prior to filtering these matches with the subsequent convolutional layers and selecting peaks corresponding to the ground-truth ECG. The latent representation of R-peak information is then fed into a R-peak classifier, of which the output provides precise R-peak locations. The proposed Deep Matched Filter is evaluated using leave-one-subject-out cross-validation over 36 subjects with an age range of 18-75, with the Deep-MF outperforming existing algorithms for R-peak detection in noisy ECG. The Deep-MF achieves a median R-peak recall of 94.9% and a median precision of 91.2% across subjects when evaluated with leave-one-subject-out cross validation. Overall, this Deep-Match framework serves as a valuable step forward for the real-world functionality of Ear-ECG and, through its interpretable operation, the acceptance of deep learning models in e-Health.

From Scalp to Ear-EEG: A Generalizable Transfer Learning Model for Automatic Sleep Scoring in Older People.

OBJECTIVE: Sleep monitoring has extensively utilized electroencephalogram (EEG) data collected from the scalp, yielding very large data repositories and well-trained analysis models. Yet, this wealth of data is lacking for emerging, less intrusive modalities, such as ear-EEG. METHODS AND PROCEDURES: The current study seeks to harness the abundance of open-source scalp EEG datasets by applying models pre-trained on data, either directly or with minimal fine-tuning; this is achieved in the context of effective sleep analysis from ear-EEG data that was recorded using a single in-ear electrode, referenced to the ipsilateral mastoid, and developed in-house as described in our previous work. Unlike previous studies, our research uniquely focuses on an older cohort (17 subjects aged 65-83, mean age 71.8 years, some with health conditions), and employs LightGBM for transfer learning, diverging from previous deep learning approaches. RESULTS: Results show that the initial accuracy of the pre-trained model on ear-EEG was 70.1%, but fine-tuning the model with ear-EEG data improved its classification accuracy to 73.7%. The fine-tuned model exhibited a statistically significant improvement (p < 0.05, dependent t-test) for 10 out of the 13 participants, as reflected by an enhanced average Cohen's kappa score (a statistical measure of inter-rater agreement for categorical items) of 0.639, indicating a stronger agreement between automated and expert classifications of sleep stages. Comparative SHAP value analysis revealed a shift in feature importance for the N3 sleep stage, underscoring the effectiveness of the fine-tuning process. CONCLUSION: Our findings underscore the potential of fine-tuning pre-trained scalp EEG models on ear-EEG data to enhance classification accuracy, particularly within an older population and using feature-based methods for transfer learning. This approach presents a promising avenue for ear-EEG analysis in sleep studies, offering new insights into the applicability of transfer learning across different populations and computational techniques. CLINICAL IMPACT: An enhanced ear-EEG method could be pivotal in remote monitoring settings, allowing for continuous, non-invasive sleep quality assessment in elderly patients with conditions like dementia or sleep apnea.

Hearables: feasibility of recording cardiac rhythms from single in-ear locations.

The ear is well positioned to accommodate both brain and vital signs monitoring, via so-called hearable devices. Consequently, ear-based electroencephalography has recently garnered great interest. However, despite the considerable potential of hearable based cardiac monitoring, the biophysics and characteristic cardiac rhythm of ear-based electrocardiography (ECG) are not yet well understood. To this end, we map the cardiac potential on the ear through volume conductor modelling and measurements on multiple subjects. In addition, in order to demonstrate real-world feasibility of in-ear ECG, measurements are conducted throughout a long-time simulated driving task. As a means of evaluation, the correspondence between the cardiac rhythms obtained via the ear-based and standard Lead I measurements, with respect to the shape and timing of the cardiac rhythm, is verified through three measures of similarity: the Pearson correlation, and measures of amplitude and timing deviations. A high correspondence between the cardiac rhythms obtained via the ear-based and Lead I measurements is rigorously confirmed through agreement between simulation and measurement, while the real-world feasibility was conclusively demonstrated through efficacious cardiac rhythm monitoring during prolonged driving. This work opens new avenues for seamless, hearable-based cardiac monitoring that extends beyond heart rate detection to offer cardiac rhythm examination in the community.

Physically Meaningful Surrogate Data for COPD.

The rapidly increasing prevalence of debilitating breathing disorders, such as chronic obstructive pulmonary disease (COPD), calls for a meaningful integration of artificial intelligence (AI) into respiratory healthcare. Deep learning techniques are "data hungry" whilst patient-based data is invariably expensive and time consuming to record. To this end, we introduce a novel COPD-simulator, a physical apparatus with an easy to replicate design which enables rapid and effective generation of a wide range of COPD-like data from healthy subjects, for enhanced training of deep learning frameworks. To ensure the faithfulness of our domain-aware COPD surrogates, the generated waveforms are examined through both flow waveforms and photoplethysmography (PPG) waveforms (as a proxy for intrathoracic pressure) in terms of duty cycle, sample entropy, FEV1/FVC ratios and flow-volume loops. The proposed simulator operates on healthy subjects and is able to generate FEV1/FVC obstruction ratios ranging from greater than 0.8 to less than 0.2, mirroring values that can observed in the full spectrum of real-world COPD. As a final stage of verification, a simple convolutional neural network is trained on surrogate data alone, and is used to accurately detect COPD in real-world patients. When training solely on surrogate data, and testing on real-world data, a comparison of true positive rate against false positive rate yields an area under the curve of 0.75, compared with 0.63 when training solely on real-world data.

Hearables: Heart Rate Variability from Ear Electrocardiogram and Ear Photoplethysmogram (Ear-ECG and Ear-PPG).

This work aims to classify physiological states using heart rate variability (HRV) features extracted from electrocardiograms recorded in the ears (ear-ECG). The physiological states considered in this work are: (a) normal breathing, (b) controlled slow breathing, and (c) mental exercises. Since both (b) and (c) cause higher variance in heartbeat intervals, breathing-related features (SpO2 and mean breathing interval) from the ear Photoplethysmogram (ear-PPG) are used to facilitate classification. This work: 1) proposes a scheme that, after initialization, automatically extracts R-peaks from low signal-to-noise ratio ear-ECG; 2) verifies the feasibility of extracting meaningful HRV features from ear-ECG; 3) quantitatively compares several ear-ECG sites; and 4) discusses the benefits of combining ear-ECG and ear-PPG features.

Reducing racial bias in SpO2 estimation: The effects of skin pigmentation.

Accurate pulse-oximeter readings are critical for clinical decisions, especially when arterial blood-gas tests - the gold standard for determining oxygen saturation levels - are not available, such as when determining COVID-19 severity. Several studies demonstrate that pulse oxygen saturation estimated from photoplethysmography (PPG) introduces a racial bias due to the more profound scattering of light in subjects with darker skin due to the increased presence of melanin. This leads to an overestimation of blood oxygen saturation in those with darker skin that is increased for low blood oxygen levels and can result in a patient not receiving potentially life-saving supplemental oxygen. This racial bias has been comprehensively studied in conventional finger pulse oximetry but in other less commonly used measurement sites, such as in-ear pulse oximetry, it remains unexplored. Different measurement sites can have thinner epidermis compared with the finger and lower exposure to sunlight (such as is the case with the ear canal), and we hypothesise that this could reduce the bias introduced by skin tone on pulse oximetry. To this end, we compute SpO2 in different body locations, during rest and breath-holds, and compare with the index finger. The study involves a participant pool covering 6-pigmentation categories from Fitzpatrick's Skin Pigmentation scale. These preliminary results indicate that locations characterized by cartilaginous highly vascularized tissues may be less prone to the influence of melanin and pigmentation in the estimation of SpO2, paving the way for the development of non-discriminatory pulse oximetry devices.

A class of doubly stochastic shift operators for random graph signals and their boundedness.

A class of doubly stochastic graph shift operators (GSO) is proposed, which is shown to exhibit: (i) lower and upper L2-boundedness for locally stationary random graph signals, (ii) L2-isometry for i.i.d. random graph signals with the asymptotic increase in the incoming neighbourhood size of vertices, and (iii) preservation of the mean of any graph signal - all prerequisites for reliable graph neural networks. These properties are obtained through a statistical consistency analysis of the proposed graph shift operator, and by exploiting the dual role of the doubly stochastic GSO as a Markov (diffusion) matrix and as an unbiased expectation operator. For generality, we consider directed graphs which exhibit asymmetric connectivity matrices. The proposed approach is validated through an example on the estimation of a vector field.

Tensor decomposition and machine learning for the detection of arteriovenous fistula stenosis: An initial evaluation.

Duplex ultrasound (DUS) is the most widely used method for surveillance of arteriovenous fistulae (AVF) created for dialysis. However, DUS is poor at predicting AVF outcomes and there is a need for novel methods that can more accurately evaluate multidirectional AVF flow. In this study we aimed to evaluate the feasibility of detecting AVF stenosis using a novel method combining tensor-decomposition of B-mode ultrasound cine loops (videos) of blood flow and machine learning classification. Classification of stenosis was based on the DUS assessment of blood flow volume, vessel diameter size, flow velocity, and spectral waveform features. Real-time B-mode cine loops of the arterial inflow, anastomosis, and venous outflow of the AVFs were analysed. Tensor decompositions were computed from both the 'full-frame' (whole-image) videos and 'cropped' videos (to include areas of blood flow only). The resulting output were labelled for the presence of stenosis, as per the DUS findings, and used as a set of features for classification using a Long Short-Term Memory (LSTM) neural network. A total of 61 out of 66 available videos were used for analysis. The whole-image classifier failed to beat random guessing, achieving a mean area under the receiver operating characteristics (AUROC) value of 0.49 (CI 0.48 to 0.50). In contrast, the 'cropped' video classifier performed better with a mean AUROC of 0.82 (CI 0.66 to 0.96), showing promising predictive power despite the small size of the dataset. The combined application of tensor decomposition and machine learning are promising for the detection of AVF stenosis and warrant further investigation.

Feasibility of Transfer Learning from Finger PPG to In-Ear PPG.

The success of deep learning methods has enabled many modern wearable health applications, but has also highlighted the critical caveat of their extremely data hungry nature. While the widely explored wrist and finger photoplethysmography (PPG) sites are less affected, given the large available databases, this issue is prohibitive to exploring the full potential of novel recording locations such as in-ear wearables. To this end, we assess the feasibility of transfer learning from finger PPG to in-ear PPG in the context of deep learning for respiratory monitoring. This is achieved by introducing an encoder-decoder framework which is set up to extract respiratory waveforms from PPG, whereby simultaneously recorded gold standard respiratory waveforms (capnography, impedance pneumography and air flow) are used as a training reference. Next, the data augmentation and training pipeline is examined for both training on finger PPG and the subsequent fine tuning on in-ear PPG. The results indicate that, through training on two large finger PPG data sets (95 subjects) and then retraining on our own small in-ear PPG data set (6 subjects), the model achieves lower and more consistent test error for the prediction of the respiratory waveforms, compared to training on the small in-ear data set alone. This conclusively demonstrates the feasibility of transfer learning from finger PPG to in-ear PPG, leading to better generalisation across a wide range of respiratory rates.