Chinese Immigrant Caregivers: Understanding Their Unmet Needs and the Co-Design of an mHealth App.

BACKGROUND: Immigrant caregivers support the aging population, yet their own needs are often neglected. Mobile technology-facilitated interventions can promote caregiver health by providing easy access to self-care materials. OBJECTIVE: This study employed a design thinking framework to examine Chinese immigrant caregivers' (CICs) unmet self-care needs and co-design an app for promoting self-care with CICs. METHODS: Nineteen semi-structured interviews were conducted in conceptual design and prototype co-design phases. FINDINGS: Participants reported unmet self-care needs influenced by psychological and social barriers, immigrant status, and caregiving tasks. They expressed the need to learn to keep healthy boundaries with the care recipient and respond to emergencies. Gaining knowledge was the main benefit that drew CICs' interest in using the self-care app. However, potential barriers to use included issues of curriculum design, technology anxiety, limited free time, and caregiving burdens. DISCUSSION: The co-design process appears to be beneficial in having participants voice both barriers and preferences.

Construction and Verification of Predictive Model for Influencing Factors of Quality of Life in Patients with Type 2 Diabetic Nephropathy: A Hospital-Based Retrospective Study.

OBJECTIVE: The influencing factors of quality of life (QOL) in patients with type 2 diabetic nephropathy (T2DN) were explored, a practical risk prediction model was constructed and independent verification was conducted. METHODS: The clinical data of 273 patients with T2DN in Tai'an Maternal and Child Health Care Center from February 2021 to February 2023 were used for retrospective analysis, and the patients were divided into modelling group (n = 173) and validation group (n = 100). According to 36-item short form health survey (SF-36) scores, the research subjects in the modelling group were divided further into poor group (n = 78) and good group (n = 95). Multivariate logistic regression was used in analysing the influencing factors of QOL and establishing a clinical prediction model based on the results. Then, a receiver operating characteristic (ROC) curve was used in evaluating the model's prediction efficiency. RESULTS: Remarkable differences in age, duration of diabetes, presence or absence of hypertension, education level, exercise frequency and family monthly income were found among the patients (p < 0.001). Multivariate logistic regression analysis suggested age ≥60 (odds ratio (OR) = 3.395, 95% CI = 1.269-9.083), duration of diabetes ≥3 years (OR = 4.574, 95% CI = 1.623-12.885), presence of hypertension (OR = 4.011, 95% CI = 1.490-10.796), education level of junior high school and below (OR = 7.318, 95% CI = 3.648-14.678), no or little exercise (OR = 3.948, 95% CI = 1.989-7.839) and family monthly income <3500 yuan (OR = 2.871, 95% CI = 1.089-7.573) are risk factors for poor QOL (p < 0.05). The regression model was logit (p) = -5.412 + 1.222X1 + 1.520X2 + 1.389X3 + 1.990X4 + 1.373X5 + 1.055X6 (X1 as age ≥60, X2 as duration of diabetes ≥3 years, X3 as presence of hypertension, X4 as education level of below junior high school, X5 as no or little exercise and X6 as family monthly income <3500 yuan). Based on this model, the ROC curve showed that the AUC value, standard error and 95% CI were 0.842, 0.043 and 0.758-0.926, respectively. An analysis was made on the re-included 100 patients, and the predictive sensitivity, specificity and Kappa coefficient of the constructed model were 82.10%, 86.90% and 0.703. CONCLUSIONS: Age ≥60, duration of diabetes ≥3 years, presence of hypertension, education level of junior high school and below, no or little exercise and family monthly income <3500 yuan were independent influencing factors for poor QOL in patients with T2DN. The use of this model has certain clinical application value.

Glance and Focus Networks for Dynamic Visual Recognition.

Spatial redundancy widely exists in visual recognition tasks, i.e., discriminative features in an image or video frame usually correspond to only a subset of pixels, while the remaining regions are irrelevant to the task at hand. Therefore, static models which process all the pixels with an equal amount of computation result in considerable redundancy in terms of time and space consumption. In this paper, we formulate the image recognition problem as a sequential coarse-to-fine feature learning process, mimicking the human visual system. Specifically, the proposed Glance and Focus Network (GFNet) first extracts a quick global representation of the input image at a low resolution scale, and then strategically attends to a series of salient (small) regions to learn finer features. The sequential process naturally facilitates adaptive inference at test time, as it can be terminated once the model is sufficiently confident about its prediction, avoiding further redundant computation. It is worth noting that the problem of locating discriminant regions in our model is formulated as a reinforcement learning task, thus requiring no additional manual annotations other than classification labels. GFNet is general and flexible as it is compatible with any off-the-shelf backbone models (such as MobileNets, EfficientNets and TSM), which can be conveniently deployed as the feature extractor. Extensive experiments on a variety of image classification and video recognition tasks and with various backbone models demonstrate the remarkable efficiency of our method. For example, it reduces the average latency of the highly efficient MobileNet-V3 on an iPhone XS Max by 1.3x without sacrificing accuracy. Code and pre-trained models are available at https://github.com/blackfeather-wang/GFNet-Pytorch.

Spatially Adaptive Feature Refinement for Efficient Inference.

Spatial redundancy commonly exists in the learned representations of convolutional neural networks (CNNs), leading to unnecessary computation on high-resolution features. In this paper, we propose a novel Spatially Adaptive feature Refinement (SAR) approach to reduce such superfluous computation. It performs efficient inference by adaptively fusing information from two branches: one conducts standard convolution on input features at a lower spatial resolution, and the other one selectively refines a set of regions at the original resolution. The two branches complement each other in feature learning, and both of them evoke much less computation than standard convolution. SAR is a flexible method that can be conveniently plugged into existing CNNs to establish models with reduced spatial redundancy. Experiments on CIFAR and ImageNet classification, COCO object detection and PASCAL VOC semantic segmentation tasks validate that the proposed SAR can consistently improve the network performance and efficiency. Notably, our results show that SAR only refines less than 40% of the regions in the feature representations of a ResNet for 97% of the samples in the validation set of ImageNet to achieve comparable accuracy with the original model, revealing the high computational redundancy in the spatial dimension of CNNs.

Detecting Coal Pulverizing System Anomaly Using a Gated Recurrent Unit and Clustering.

The coal pulverizing system is an important auxiliary system in thermal power generation systems. The working condition of a coal pulverizing system may directly affect the safety and economy of power generation. Prognostics and health management is an effective approach to ensure the reliability of coal pulverizing systems. As the coal pulverizing system is a typical dynamic and nonlinear high-dimensional system, it is difficult to construct accurate mathematical models used for anomaly detection. In this paper, a novel data-driven integrated framework for anomaly detection of the coal pulverizing system is proposed. A neural network model based on gated recurrent unit (GRU) networks, a type of recurrent neural network (RNN), is constructed to describe the temporal characteristics of high-dimensional data and predict the system condition value. Then, aiming at the prediction error, a novel unsupervised clustering algorithm for anomaly detection is proposed. The proposed framework is validated by a real case study from an industrial coal pulverizing system. The results show that the proposed framework can detect the anomaly successfully.

Noninvasive prenatal paternity testing by maternal plasma DNA sequencing in twin pregnancies.

SNPs, combined with massively parallel sequencing technology, have proven applicability in noninvasive prenatal paternity testing (NIPPT) for singleton pregnancies in our previous research, using circulating cell-free DNA in maternal plasma. However, the feasibility of NIPPT in twin pregnancies has remained uncertain. As a pilot study, we developed a practical method to noninvasively determine the paternity of twin pregnancies by maternal plasma DNA sequencing based on a massively parallel sequencing platform. Blood samples were collected from 15 pregnant women (twin pregnancies at 9-18 weeks of gestation). Parental DNA and maternal plasma cell-free DNA were analyzed with custom-designed probes covering 5226 polymorphic SNP loci. A mathematical model for data interpretation was established, including the zygosity determination and paternity index calculations. Each plasma sample was independently tested against the alleged father and 90 unrelated males. As a result, the zygosity in each twin case was correctly determined, prior to paternity analysis. Further, the correct biological father was successfully identified, and the paternity of all 90 unrelated males was excluded in each case. Our study demonstrates that NIPPT can be performed for twin pregnancies. This finding may contribute to development in NIPPT and diagnosis of certain genetic diseases.

Noninvasive Prenatal Paternity Testing (NIPAT) through Maternal Plasma DNA Sequencing: A Pilot Study.

Short tandem repeats (STRs) and single nucleotide polymorphisms (SNPs) have been already used to perform noninvasive prenatal paternity testing from maternal plasma DNA. The frequently used technologies were PCR followed by capillary electrophoresis and SNP typing array, respectively. Here, we developed a noninvasive prenatal paternity testing (NIPAT) based on SNP typing with maternal plasma DNA sequencing. We evaluated the influence factors (minor allele frequency (MAF), the number of total SNP, fetal fraction and effective sequencing depth) and designed three different selective SNP panels in order to verify the performance in clinical cases. Combining targeted deep sequencing of selective SNP and informative bioinformatics pipeline, we calculated the combined paternity index (CPI) of 17 cases to determine paternity. Sequencing-based NIPAT results fully agreed with invasive prenatal paternity test using STR multiplex system. Our study here proved that the maternal plasma DNA sequencing-based technology is feasible and accurate in determining paternity, which may provide an alternative in forensic application in the future.

An Advanced Model to Precisely Estimate the Cell-Free Fetal DNA Concentration in Maternal Plasma.

BACKGROUND: With the speedy development of sequencing technologies, noninvasive prenatal testing (NIPT) has been widely applied in clinical practice for testing for fetal aneuploidy. The cell-free fetal DNA (cffDNA) concentration in maternal plasma is the most critical parameter for this technology because it affects the accuracy of NIPT-based sequencing for fetal trisomies 21, 18 and 13. Several approaches have been developed to calculate the cffDNA fraction of the total cell-free DNA in the maternal plasma. However, most approaches depend on specific single nucleotide polymorphism (SNP) allele information or are restricted to male fetuses. METHODS: In this study, we present an innovative method to accurately deduce the concentration of the cffDNA fraction using only maternal plasma DNA. SNPs were classified into four maternal-fetal genotype combinations and three boundaries were added to capture effective SNP loci in which the mother was homozygous and the fetus was heterozygous. The median value of the concentration of the fetal DNA fraction was estimated using the effective SNPs. A depth-bias correction was performed using simulated data and corresponding regression equations for adjustments when the depth of the sequencing data was below 100-fold or the cffDNA fraction is less than 10%. RESULTS: Using our approach, the median of the relative bias was 0.4% in 18 maternal plasma samples with a median sequencing depth of 125-fold. There was a significant association (r = 0.935) between our estimations and the estimations inferred from the Y chromosome. Furthermore, this approach could precisely estimate a cffDNA fraction as low as 3%, using only maternal plasma DNA at the targeted region with a sequencing depth of 65-fold. We also used PCR instead of parallel sequencing to calculate the cffDNA fraction. There was a significant association (r = 98.2%) between our estimations and those inferred from the Y chromosome.

Performance comparison between rapid sequencing platforms for ultra-low coverage sequencing strategy.

Ultra-low coverage sequencing (ULCS) is one of the most promising strategies for sequencing based clinical application. These clinical applications, especially prenatal diagnosis, have a strict requirement of turn-around-time; therefore, the application of ULCS is restricted by current high throughput sequencing platforms. Recently, the emergence of rapid sequencing platforms, such as MiSeq and Ion Proton, brings ULCS strategy into a new era. The comparison of their performance could shed lights on their potential application in large-scale clinic trials. In this study, we performed ULCS (<0.1X coverage) on both MiSeq and Ion Proton platforms for 18 spontaneous abortion fetuses carrying aneuploidy and compared their performance on different levels. Overall basic data and GC bias showed no significant difference between these two platforms. We also found the sex and aneuploidy detection indicated 100% sensitivity and 100% specificity on both platforms. Our study generated essential data from these two rapid sequencing platforms, which provides useful reference for later research and potentially accelerates the clinical applications of ULCS.

PSCC: sensitive and reliable population-scale copy number variation detection method based on low coverage sequencing.

BACKGROUND: Copy number variations (CNVs) represent an important type of genetic variation that deeply impact phenotypic polymorphisms and human diseases. The advent of high-throughput sequencing technologies provides an opportunity to revolutionize the discovery of CNVs and to explore their relationship with diseases. However, most of the existing methods depend on sequencing depth and show instability with low sequence coverage. In this study, using low coverage whole-genome sequencing (LCS) we have developed an effective population-scale CNV calling (PSCC) method. METHODOLOGY/PRINCIPAL FINDINGS: In our novel method, two-step correction was used to remove biases caused by local GC content and complex genomic characteristics. We chose a binary segmentation method to locate CNV segments and designed combined statistics tests to ensure the stable performance of the false positive control. The simulation data showed that our PSCC method could achieve 99.7%/100% and 98.6%/100% sensitivity and specificity for over 300 kb CNV calling in the condition of LCS (∼2×) and ultra LCS (∼0.2×), respectively. Finally, we applied this novel method to analyze 34 clinical samples with an average of 2× LCS. In the final results, all the 31 pathogenic CNVs identified by aCGH were successfully detected. In addition, the performance comparison revealed that our method had significant advantages over existing methods using ultra LCS. CONCLUSIONS/SIGNIFICANCE: Our study showed that PSCC can sensitively and reliably detect CNVs using low coverage or even ultra-low coverage data through population-scale sequencing.

Noninvasive prenatal detection for pathogenic CNVs: the application in α-thalassemia.

BACKGROUND: The discovery of cell free fetal DNA (cff-DNA) in maternal plasma has brought new insight for noninvasive prenatal diagnosis. Combining with the rapidly developed massively parallel sequencing technology, noninvasive prenatal detection of chromosome aneuploidy and single base variation has been successfully validated. However, few studies discussed the possibility of noninvasive pathogenic CNVs detection. METHODOLOGY/PRINCIPAL FINDINGS: A novel algorithm for noninvasive prenatal detection of fetal pathogenic CNVs was firstly tested in 5 pairs of parents with heterozygote α-thalassemia of Southeast Asian (SEA) deletion using target region capture sequencing for maternal plasma. Capture probes were designed for α-globin (HBA) and ß-globin (HBB) gene, as well as 4,525 SNPs selected from 22 automatic chromosomes. Mixed adaptors with 384 different barcodes were employed to construct maternal plasma DNA library for massively parallel sequencing. The signal of fetal CNVs was calculated using the relative copy ratio (RCR) of maternal plasma combined with the analysis of R-score and L-score by comparing with normal control. With mean of 101.93× maternal plasma sequencing depth for the target region, the RCR value combined with further R-score and L-score analysis showed a possible homozygous deletion in the HBA gene region for one fetus, heterozygous deletion for two fetus and normal for the other two fetus, which was consistent with that of invasive prenatal diagnosis. CONCLUSIONS/SIGNIFICANCE: Our study showed the feasibility to detect pathogenic CNVs using target region capture sequencing, which might greatly extend the scope of noninvasive prenatal diagnosis.

Effective noninvasive zygosity determination by maternal plasma target region sequencing.

BACKGROUND: Currently very few noninvasive molecular genetic approaches are available to determine zygosity for twin pregnancies in clinical laboratories. This study aimed to develop a novel method to determine zygosity by using maternal plasma target region sequencing. METHODS: We constructed a statistic model to calculate the possibility of each zygosity type using likelihood ratios ( Li ) and empirical dynamic thresholds targeting at 4,524 single nucleotide polymorphisms (SNPs) loci on 22 autosomes. Then two dizygotic (DZ) twin pregnancies,two monozygotic (MZ) twin pregnancies and two singletons were recruited to evaluate the performance of our novel method. Finally we estimated the sensitivity and specificity of the model in silico under different cell-free fetal DNA (cff-DNA) concentration and sequence depth. RESULTS/CONCLUSIONS: We obtained 8.90 Gbp sequencing data on average for six clinical samples. Two samples were classified as DZ with L values of 1.891 and 1.554, higher than the dynamic DZ cut-off values of 1.162 and 1.172, respectively. Another two samples were judged as MZ with 0.763 and 0.784 of L values, lower than the MZ cut-off values of 0.903 and 0.918. And the rest two singleton samples were regarded as MZ twins, with L values of 0.639 and 0.757, lower than the MZ cut-off values of 0.921 and 0.799. In silico, the estimated sensitivity of our noninvasive zygosity determination was 99.90% under 10% total cff-DNA concentration with 2 Gbp sequence data. As the cff-DNA concentration increased to 15%, the specificity was as high as 97% with 3.50 Gbp sequence data, much higher than 80% with 10% cff-DNA concentration. SIGNIFICANCE: This study presents the feasibility to noninvasively determine zygosity of twin pregnancy using target region sequencing, and illustrates the sensitivity and specificity under various detecting condition. Our method can act as an alternative approach for zygosity determination of twin pregnancies in clinical practice.