The security, privacy, and safety issues around Female-oriented technologies (FemTech) and data can lead to differential harms. These complex risks and harms are enabled by many factors including inadequate regulations, the non-compliant practices of the industry, and the lack of research and guidelines for cyber-secure, privacy-preserving, and safe products. In this paper, we review the existing regulations related to FemTech in the United Kingdom, EU, and Switzerland and identify the gaps. We run experiments on a range of FemTech devices and apps and identify several exploitative practices. We advocate for the policymakers to explicitly acknowledge and accommodate the risks of these technologies in the relevant regulations.

INTRODUCTION:The ability to collect blood samples remotely without the involvement of healthcare professionals is a key element of future telehealth applications. We developed and validated the application of the Drawbridge OneDraw device for use at home for blood sample collection. The device was then applied in a large population-based remote monitoring study to assess changes in SARS-CoV-2 IgG antibody levels.METHODS:We tested: (1) feasibility of participants using the device at home without a healthcare professional on the upper arm and thigh sites (2) stability of the dried blood sample collected remotely (3) participant acceptability of the device compared with finger-prick and venous blood samples and the validity of SARS-CoV-2 virus antibody measurement versus venous blood sample (4) application to the Fenland COVID-19 study in which 4023 participants at 3 timepoints across 6 months.RESULTS:Participant acceptability was high, with a significantly lower median perceived pain score and 76% of participants preferring the OneDraw device over the other blood collection methods. There was high level of agreement in SARS-CoV-2 virus antibody results with venous blood samples in 120 participants (Cohen's kappa 0.68 (95% CI 0.56, 0.83). In the Fenland COVID-19 study, 92% of participants returned a sample at baseline (3702/4023), 89% at 3 months (3492/3918) and 93% at 6 months (3453/3731), with almost all samples received successfully processed (99.9%).DISCUSSION:The OneDraw device enables a standardised blood sample collection at home by participants themselves. Due to its ease-of-use and acceptability the OneDraw device is particularly useful in telehealth approaches where multiple samples need to be collected.

Transformers have rapidly become the dominant architecture for analyzing sequential data, utilizing their self-attention mechanism to effectively capture long-term temporal patterns, outperforming recurrent-based methods across various applications. In this paper, we explore the application of transformers to wearable sensor data, focusing on the analysis of human gait, which is often complex and sensitive. We propose two novel frameworks: Data Efficient Sensor Transformer (DesT) for centralized learning and Federated Data Efficient Sensor Transformer (FeDesT) for federated learning (FL) in edge-computing environments. Both frameworks employ knowledge distillation to improve the generalization of transformers, which can be prone to over-fitting due to the limited labeled data available in wearable sensor applications. Experimental results using human gait data collected from uneven and irregular surfaces show that DesT improves the accuracy by 14.8% when compared to existing transformers. FeDesT reduces computational demands on edge devices while outperforming traditional FL methods for transformers. This work demonstrates the potential of transformers for wearable sensor data analysis in both centralized and federated contexts, particularly where privacy and computational efficiency is paramount.

Aims: As sustained weight loss is vital for achieving remission of type 2 diabetes, we explored whether randomisation to weight loss plus maintenance in the DiRECT trial was associated with physical activity, inactivity or sleep. Methods: Participants were randomised to either a dietary weight management programme or best-practice care. The weight management group were encouraged to increase daily physical activity to their sustainable maximum. Objective measurement was achieved using a wrist-worn GENEActiv accelerometer for 7 days at baseline, 12 and 24 months in both groups. Results: Despite average weight loss of 10 kg at 12 months in the intervention (n = 66) group, there were no differences in total physical activity or inactivity compared with the control (n = 104) at any time point. However, in our exploratory analysis, those who lost more than 10% of their baseline body weight performed on average 11 mins/day more light activity than the < 10% group at 24 months (p = 0.033) and had significantly lower bouts of Inactivity(30min) (interaction, p = 0.005) across 12 and 24 months. At 24 months, the >= 10% group had higher daily acceleration (38.5 +/- 12.1 vs. 33.2 +/- 11.1 mg, p = 0.020), and higher accelerations in the most active 5-hour period (59.4 +/- 21.8 vs. 50.6 +/- 18.3 mg, p = 0.023). Wakefulness after sleep onset decreased in the intervention group compared with the control group and also in the & GE;10% weight loss group at 12 and 24 months. Conclusions: Randomisation to a successful intensive weight loss intervention, including regular physical activity encouragement, was not associated with increased physical activity although sleep parameters improved. Physical activity was greater, and night-time waking reduced in those who maintained > 10% weight loss at 12 and 24 months.

This study introduces the Data Efficient Separable Transformer (DeSepTr) architecture, a novel framework for Human Activity Recognition (HAR) that utilizes a light-weight computer vision model to train a Vision Transformer (ViT) on spectrograms generated from wearable sensor data. The proposed model achieves strong results on several HAR tasks, including surface condition recognition and activity recognition. Compared to the ResNet-18 model, DeSepTr outperforms by 5.9% on out-of-distribution test data accuracy for surface condition recognition. The framework enables ViTs to learn from limited labeled training data and generalize to data from participants outside of the training cohort, potentially leading to the development of activity recognition models that are robust to the wider population. The results suggest that the DeSepTr architecture can overcome limitations related to the heterogeneity of individuals’ behavior patterns and the weak inductive bias of transformer algorithms.

Primary care EHR data are often of clinical importance to cohort studies however they require careful handling. Challenges include determining the periods during which EHR data were collected. Participants are typically censored when they deregister from a medical practice, however, cohort studies wish to follow participants longitudinally including those that change practice. Using UK Biobank as an exemplar, we developed methodology to infer continuous periods of data collection and maximize follow-up in longitudinal studies. This resulted in longer follow-up for around 40% of participants with multiple registration records (mean increase of 3.8 years from the first study visit). The approach did not sacrifice phenotyping accuracy when comparing agreement between self-reported and EHR data. A diabetes mellitus case study illustrates how the algorithm supports longitudinal study design and provides further validation. We use UK Biobank data, however, the tools provided can be used for other conditions and studies with minimal alteration.

Physical activity (PA) can reduce cardiovascular disease (CVD) risk factors, and although primary care settings offer a large reach to promote PA and reduce CVD risk, primary health care professionals may lack self-efficacy and tools to effectively promote PA in practice. Movement as Medicine for CVD Prevention is a suite of 2 theory-based, web-based behavioral interventions—one for health care professionals and one for patients—which may offer a pathway for promoting PA and reducing CVD risk in primary care. This study aims to examine the feasibility and possible effects of Movement as Medicine for CVD Prevention. This nonrandomized pilot study recruited participants from primary care organizations in the Northeast of England. Enrolled health care professionals followed a theory-based, web-based course on PA counseling and motivational interviewing techniques. After the course, health care professionals delivered behavior change consultations based on motivational interviewing to inactive individuals with >20% risk of developing CVD within 10 years. Patients were then given access to a website based on self-determination and self-regulation theories, which targeted increased levels of PA. Outcomes were assessed at baseline and after 3 months, and patient data were analyzed on an intention-to-treat basis in a multiple imputation data set. Recruitment rates of primary care organizations fell below expectations. A total of 11 health care professionals from 3 enrolled primary care organizations completed the web-based course and reported increases in important theoretical determinants of PA promotion in practice (eg, self-efficacy, Cohen d=1.24, 95% CI 0.67-1.80; and planning, Cohen d=0.85, 95% CI −0.01 to 1.69). A total of 83 patients were enrolled in the study, and 58 (70%) completed both the baseline and 3-month assessments. Compared with baseline, patients had higher levels of objective (Cohen d=0.77, 95% CI 0.13-1.41) but not subjective (Cohen d=0.40, 95% CI −0.03 to 0.83) moderate to vigorous PA at 3 months. Patients also reported higher levels of the PA determinants of intention, self-efficacy, intrinsic motivation, and action planning and action control at 3 months (effect sizes ranged from Cohen d=0.39 to 0.60). The Movement as Medicine for CVD Prevention intervention seems to have the potential to improve patient PA behaviors and important determinants of health care professionals’ PA promotion practices. However, the recruitment rates of primary care organizations in this study were low and would need to be increased to examine the efficacy of the program. This study offers several insights into improving the feasibility of this primary care PA promotion pathway. ISRCTN Registry ISRCTN14582348; http://www.isrctn.com/ISRCTN14582348

Sleep is a fundamental physiological process that is essential for sustaining a healthy body and mind. The gold standard for clinical sleep monitoring is polysomnography(PSG), based on which sleep can be categorized into five stages, including wake/rapid eye movement sleep (REM sleep)/Non-REM sleep 1 (N1)/Non-REM sleep 2 (N2)/Non-REM sleep 3 (N3). However, PSG is expensive, burdensome and not suitable for daily use. For long-term sleep monitoring, ubiquitous sensing may be a solution. Most recently, cardiac and movement sensing has become popular in classifying three-stage sleep, since both modalities can be easily acquired from research-grade or consumer-grade devices (e.g., Apple Watch). However, how best to fuse the data for greatest accuracy remains an open question. In this work, we comprehensively studied deep learning (DL)-based advanced fusion techniques consisting of three fusion strategies alongside three fusion methods for three-stage sleep classification based on two publicly available datasets. Experimental results demonstrate important evidences that three-stage sleep can be reliably classified by fusing cardiac/movement sensing modalities, which may potentially become a practical tool to conduct large-scale sleep stage assessment studies or long-term self-tracking on sleep. To accelerate the progression of sleep research in the ubiquitous/wearable computing community, we made this project open source, and the code can be found at: https://github.com/bzhai/Ubi-SleepNet.

BackgroundBetween 2013 and 2015, the UK Biobank collected accelerometer traces from 103,712 volunteers aged between 40 and 69 years using wrist-worn triaxial accelerometers for 1 week. This data set has been used in the past to verify that individuals with chronic diseases exhibit reduced activity levels compared with healthy populations. However, the data set is likely to be noisy, as the devices were allocated to participants without a set of inclusion criteria, and the traces reflect free-living conditions. ObjectiveThis study aims to determine the extent to which accelerometer traces can be used to distinguish individuals with type 2 diabetes (T2D) from normoglycemic controls and to quantify their limitations. MethodsMachine learning classifiers were trained using different feature sets to segregate individuals with T2D from normoglycemic individuals. Multiple criteria, based on a combination of self-assessment UK Biobank variables and primary care health records linked to UK Biobank participants, were used to identify 3103 individuals with T2D in this population. The remaining nondiabetic 19,852 participants were further scored on their physical activity impairment severity based on other conditions found in their primary care data, and those deemed likely physically impaired at the time were excluded. Physical activity features were first extracted from the raw accelerometer traces data set for each participant using an algorithm that extends the previously developed Biobank Accelerometry Analysis toolkit from Oxford University. These features were complemented by a selected collection of sociodemographic and lifestyle features available from UK Biobank. ResultsWe tested 3 types of classifiers, with an area under the receiver operating characteristic curve (AUC) close to 0.86 (95% CI 0.85-0.87) for all 3 classifiers and F1 scores in the range of 0.80-0.82 for T2D-positive individuals and 0.73-0.74 for T2D-negative controls. Results obtained using nonphysically impaired controls were compared with highly physically impaired controls to test the hypothesis that nondiabetic conditions reduce classifier performance. Models built using a training set that included highly impaired controls with other conditions had worse performance (AUC 0.75-0.77; 95% CI 0.74-0.78; F1 scores in the range of 0.76-0.77 for T2D positives and 0.63-0.65 for controls). ConclusionsGranular measures of free-living physical activity can be used to successfully train machine learning models that are able to discriminate between individuals with T2D and normoglycemic controls, although with limitations because of the intrinsic noise in the data sets. From a broader clinical perspective, these findings motivate further research into the use of physical activity traces as a means of screening individuals at risk of diabetes and for early detection, in conjunction with routinely used risk scores, provided that appropriate quality control is enforced on the data collection protocol.

Thirty three pairs of chest radiographs taken up to 10 years apart were obtained for 33 subjects suffering from asbestosis who had applied for compensation to the Pneumoconiosis Medical Board of Western Australia. Multiple films from the period before the first radiograph in each pair, from the intervening period between the two, and from the period subsequent to the second radiograph were also available and all films were read by two independent readers according to the 1980 ILO classification of pneumoconiosis. Films were read twice as side by side pairs ten years apart, twice as two separate randomly ordered films ten years apart, and once as part of the full series of all available chest radiographs on each subject to assess which method provided the best consistency (between reader variation) and repeatability (within reader variation). Judging by consistency, the full series method performed as well as either of the other methods when assessing radiographic changes and significantly better when assessing the level of profusion of small opacities. There was little to choose between the other two methods either judging by consistency or repeatability, which could not be estimated for the full series method. Use of all available films for a subject is recommended for assessing single films, as in a prevalence study, as well as for documenting change in a longitudinal study.

Gait data collected using wearable sensors offers non-intrusive, affordable, real-time monitoring of human motion. Recognizing surface conditions from wearable sensor data has the potential to help systems discriminate between ‘poor quality’ walking data. This research investigates the predictive capabilities of machine learning models, trained on both centralized and decentralized datasets, at categorizing uneven and irregular surface conditions. The results showed that machine learning classification algorithms, trained with data originating from a single sensor positioned on the left-shank, were able to accurately discriminate between different types of surface conditions. We found the Support Vector Machine, when trained with the data centralized, had a test-set accuracy of 94%. Federated Learning offers a way to increase privacy and security for healthcare applications by avoiding the centralization of data. Our simulated federated Deep Neural Network converged to a test-accuracy of 85%, which was 8% less than the centralized counterpart.

1Faculty of Medical Sciences, Translational and Clinical Research Institute, Newcastle University, Newcastle Upon Tyne, UK; 2Central Clinical School, Faculty of Medicine and Health, The University of Sydney, Sydney, NSW, Australia; 3National Innovation Centre for Data, School of Computing, The Catalyst, Newcastle Helix, Newcastle Upon Tyne, UK; 4NIHR Innovation Observatory, The Catalyst, Newcastle Helix, Newcastle upon Tyne, UK; 5Colworth Science Park, Sharnbrook, Bedfordshire, UK; 6National Innovation Centre for Ageing, The Catalyst, Newcastle Helix, Newcastle upon Tyne, UK; 7Department of Neurology, Royal Victoria Infirmary, Newcastle upon Tyne, UK Purpose: Short and long sleep durations have adverse effects on physical and mental health. However, most studies are based on self-reported sleep duration and health status. Therefore, this longitudinal study aims to investigate objectively measured sleep duration and subsequent primary health care records in older adults to investigate the impact of sleep duration and fragmentation on physical and mental health. Methods: Data on objective sleep duration were measured using accelerometry. Primary care health records were then obtained from the UK Biobank (n=84,404). Participants (mean age, 62.4 years) were divided into five groups according to their sleep duration derived from the accelerometry data: <5 hours, 5–6 hours, 6–7 hours, 7–8 hours and >8 hours. ICD-10 codes were used for the analysis of primary care data. Wake after sleep onset, activity level during the least active 5 hours and episodes of movement during sleep were analysed as an indication for sleep fragmentation. Binary regression models were adjusted for age, gender and Townsend deprivation score. Results: A “U-shaped” relationship was found between sleep duration and diseases including diabetes, hypertension and heart disease and depression. Short and long sleep durations and fragmented sleep were associated with increased odds of disease. Conclusion: Six to eight hours of sleep, as well as less fragmented sleep, predicted better long-term metabolic and mental health.

Sleep is a fundamental physiological process that is essential for sustaining a healthy body and mind. The gold standard for clinical sleep monitoring is polysomnography(PSG), based on which sleep can be categorized into five stages, including wake/rapid eye movement sleep (REM sleep)/Non-REM sleep 1 (N1)/Non-REM sleep 2 (N2)/Non-REM sleep 3 (N3). However, PSG is expensive, burdensome and not suitable for daily use. For long-term sleep monitoring, ubiquitous sensing may be a solution. Most recently, cardiac and movement sensing has become popular in classifying three-stage sleep, since both modalities can be easily acquired from research-grade or consumer-grade devices (e.g., Apple Watch). However, how best to fuse the data for greatest accuracy remains an open question. In this work, we comprehensively studied deep learning (DL)-based advanced fusion techniques consisting of three fusion strategies alongside three fusion methods for three-stage sleep classification based on two publicly available datasets. Experimental results demonstrate important evidences that three-stage sleep can be reliably classified by fusing cardiac/movement sensing modalities, which may potentially become a practical tool to conduct large-scale sleep stage assessment studies or long-term self-tracking on sleep. To accelerate the progression of sleep research in the ubiquitous/wearable computing community, we made this project open source, and the code can be found at: https://github.com/bzhai/Ubi-SleepNet.

Purpose: Short and long sleep durations have adverse effects on physical and mental health. However, most studies are based on self-reported sleep duration and health status. Therefore, this longitudinal study aims to investigate objectively measured sleep duration and subsequent primary health care records in older adults to investigate the impact of sleep duration and fragmentation on physical and mental health. Methods: Data on objective sleep duration were measured using accelerometry. Primary care health records were then obtained from the UK Biobank (n=84,404). Participants (mean age, 62.4 years) were divided into five groups according to their sleep duration derived from the accelerometry data: <5 hours, 5-6 hours, 6-7 hours, 7-8 hours and >8 hours. ICD-10 codes were used for the analysis of primary care data. Wake after sleep onset, activity level during the least active 5 hours and episodes of movement during sleep were analysed as an indication for sleep fragmentation. Binary regression models were adjusted for age, gender and Townsend deprivation score. Results: A "U-shaped" relationship was found between sleep duration and diseases including diabetes, hypertension and heart disease and depression. Short and long sleep durations and fragmented sleep were associated with increased odds of disease. Conclusion: Six to eight hours of sleep, as well as less fragmented sleep, predicted better long-term metabolic and mental health.

Sleep controlled by the circadian rhythm is essential for many functions including energy conservation, memory consolidation and brain processing. Sleep duration and architecture changes with age. Sleep deprivation is very common in modern society and it has been identified as a major modifiable risk factor for many metabolic diseases. A cross-sectional analysis was carried out on baseline data from the UK Biobank (n=82995). Sociodemographic, health-related and lifestyle information were collected using touch-screen questionnaires. Sleep and physical activity parameters were measured objectively using wrist-worn accelerometers (participants were aged 43–79 years). Sleep durations have been categorised into five groups. short sleepers: (1) <5 hours/night, (2) 5–6 hours/night, (3) 6–7 hours/night; normal sleepers: (4) 7–8 hours/night; long sleepers: (5) >8 hours/night. Short objective sleep duration was associated with male gender, older age and lower social status. A greater proportion of males with a sleep duration <5 hours/night have very high risk waist circumference (>102cm) compared to normal and long sleepers (22.1%, 14.9%, 11.7%, 10.4% and 10.2%, respectively). A similar pattern was also seen in females (60.0%, 50.6% 43.9%, 41.3% and 40.6%, respectively). The percentage of participants with cardiometabolic diseases is significantly lower in those who sleep between 6–8 hours/night compared to other short and long sleepers (34.8%, 27.7%, 26.0%, 25.9% and 29.1%, respectively). They also have better health ratings and less likely to have hypertension, diabetes and cardiovascular disease. Finally, those who sleep 6–7 hours were most physically active compared to other sleep groups. In conclusion, 6–8 hours of sleep per night is associated with better metabolic health and higher physical activity level. Short sleep duration is associated with male gender and social deprivation. Although, no causal link can be established from this study, the results can help to develop interventions for targeted groups to reduce the adverse effects of poor sleep.

Study objectives Normal timing and duration of sleep is vital for all physical and mental health. However, many sleep-related studies depend on self-reported sleep measurements, which have limitations. This study aims to investigate the association of physical activity and sociodemographic characteristics including age, gender, coffee intake and social status with objective sleep measurements. Methods A cross-sectional analysis was carried out on 82995 participants within the UK Biobank cohort. Sociodemographic and lifestyle information were collected through touch-screen questionnaires in 2007-2010. Sleep and physical activity parameters were later measured objectively using wrist-worn accelerometers in 2013-2015 (participants were aged 43-79 years and wore watches for 7 days). Participants were divided into 5 groups based on their objective sleep duration per night (<5 hours, 5-6 hours, 6-7 hours, 7-8 hours and >8 hours). Binary logistic models were adjusted for age, gender and Townsend Deprivation Index. Results Participants who slept 6-7 hours/night were the most frequent (33.5%). Females had longer objective sleep duration than males. Short objective sleep duration (<6 hours) correlated with older age, social deprivation and high coffee intake. Finally, those who slept 6-7 hours/night were most physically active. Conclusions Objectively determined short sleep duration was associated with male gender, older age, low social status and high coffee intake. An inverse 'U-shaped' relationship between sleep duration and physical activity was also established. Optimal sleep duration for health in those over 60 may therefore be shorter than younger groups.

Background: Accelerometers are accurate tools to assess movement and physical activity. However, interpreting standardly used outputs is not straightforward for populations with impaired mobility. Methods: The applicability of GENEActiv was explored in a group of 30 participants with myotonic dystrophy and compared to a group of 14 healthy-controls. All participants performed a set of tests while wearing four different accelerometers (wrists and ankles): [1] standing still; [2] ten-meters walk test; [3] six-minutes walking test; and, [4] ten-meters walk/run test. Results: Relevant findings were: [1] high intra-accelerometer reliability (i.e. 0.97 to 0.99; p < 0.001); [2] each test acceleration values differ significantly between each other; [3] no inter-accelerometer reliability between wrist-worn devices and ankle-worn; and [4] a significant difference between the myotonic dystrophy group and the healthy-controls detectable at each test (i.e. Left-ankle values at six-minutes walking test: 48 +/- 17 for the myotonic dystrophy group, vs, 74 +/- 16 for the healthy-controls; p < 0.001). Conclusions: GENEActiv demonstrated to be valid and reliable, capable of detecting walking periods and discriminating different speeds. However, inter-accelerometer reliability only applied when comparing opposite sides of the same limb. Specific movement characteristics of the myotonic dystrophy group were identified and muscle strength showed not to be a full determinant of limb acceleration.

In this paper we propose a mobile optimized deep learning network based on the VGG16 architecture. Compared to the classical approach, after training has been performed the model is converted to a quantized equivalent where 32 bit floating point operations are exchanged for 8 bit ones. This reduces the strain on mobile memory and local caches while simultaneously reducing the computational complexity and energy requirement of the entire deep learning model. Aggregated testing has been performed to validate the complexity hypothesis and the quantized model has been compared to the original model in terms of accuracy. The results show that for a modest decrease in accuracy, the quantized model takes up 75% less disk space and through the 8 bit operations the computational complexity is reduced, showing a load and inference speed up of 3 – 4 times faster than the original model.

The myriad of neuropsychiatric manifestations reported in myotonic dystrophy type 1 may have its origin in alterations of complex brain network interactions at the structural level. In this study, we tested the hypothesis that altered white matter microstructural integrity and network organisation were present in a cohort of individuals with DM1 compared to unaffected controls, which was expected to be associated with CNS related disease manifestations of DM1. We performed a cross-sectional neuropsychological assessment and brain MRI in 25 myotonic dystrophy type 1 (DM1) patients and 26 age, sex and educational level matched unaffected controls. Patients were recruited from the Dutch cohort of the OPTIMISTIC study, a concluded trial which had included ambulant, genetically confirmed DM1 patients who were severely fatigued. We applied graph theoretical analysis on structural networks derived from diffusion tensor imaging (DTI) data and deterministic tractography to determine global and local network properties and performed group-wise comparisons. Furthermore, we analysed the following variables from structural MRI imaging: semi-quantitative white matter hyperintensity load andwhite matter tract integrity using tract-based spatial statistics (TBSS). Structural white matter networks in DM1 were characterised by reduced global efficiency, local efficiency and strength, while the network density was compatible to controls. Other findings included increased white matter hyperintensity load, and diffuse alterations of white matter microstructure in projection, association and commissural fibres. DTI and network measures were associated (partial correlations coefficients ranging from 0.46 to 0.55) with attention (d2 Test), motor skill (Purdue Pegboard test) and visual-constructional ability and memory (copy subtest of the Rey-Osterrieth Complex Figure Test). DTI and network measures were not associated with clinical measures of fatigue (checklist individual strength, fatigue subscale) or apathy (apathy evaluation scale - clinician version). In conclusion, our study supports the view of brain involvement in DM1 as a complex network disorder, characterised by white matter network alterations that may have relevant neuropsychological correlations. This work was supported by the European Community's Seventh Framework Programme (FP7/2007-2013; grant agreement n° 305,697) and the Marigold Foundation.