Geriatric depression and anxiety screening via deep learning using activity tracking and sleep data.

BACKGROUND: Geriatric depression and anxiety have been identified as mood disorders commonly associated with the onset of dementia. Currently, the diagnosis of geriatric depression and anxiety relies on self-reported assessments for primary screening purposes, which is uncomfortable for older adults and can be prone to misreporting. When a more precise diagnosis is needed, additional methods such as in-depth interviews or functional magnetic resonance imaging are used. However, these methods can not only be time-consuming and costly but also require systematic and cost-effective approaches. OBJECTIVE: The main objective of this study was to investigate the feasibility of training an end-to-end deep learning (DL) model by directly inputting time-series activity tracking and sleep data obtained from consumer-grade wrist-worn activity trackers to identify comorbid depression and anxiety. METHODS: To enhance accuracy, the input of the DL model consisted of step counts and sleep stages as time series data, along with minimal depression and anxiety assessment scores as non-time-series data. The basic structure of the DL model was designed to process mixed-input data and perform multi-label-based classification for depression and anxiety. Various DL models, including the convolutional neural network (CNN) and long short-term memory (LSTM), were applied to process the time-series data, and model selection was conducted by comparing the performances of the hyperparameters. RESULTS: This study achieved significant results in the multi-label classification of depression and anxiety, with a Hamming loss score of 0.0946 in the Residual Network (ResNet), by applying a mixed-input DL model based on activity tracking data. The comparison of hyper-parameter performance and the development of various DL models, such as CNN, LSTM, and ResNet contributed to the optimization of time series data processing and achievement of meaningful results. CONCLUSIONS: This study can be considered as the first to develop a mixed-input DL model based on activity tracking data for the multi-label identification of late-life depression and anxiety. The findings of the study demonstrate the feasibility and potential of using consumer-grade wrist-worn activity trackers in conjunction with DL models to improve the identification of comorbid mental health conditions in older adults. The study also established a multi-label classification framework for identifying the complex symptoms of depression and anxiety.

Online Boosting-Based Target Identification among Similar Appearance for Person-Following Robots.

It is challenging for a mobile robot to follow a specific target person in a dynamic environment, comprising people wearing similar-colored clothes and having the same or similar height. This study describes a novel framework for a person identification model that identifies a target person by merging multiple features into a single joint feature online. The proposed framework exploits the deep learning output to extract four features for tracking the target person without prior knowledge making it generalizable and more robust. A modified intersection over union between the current frame and the last frame is proposed as a feature to distinguish people, in addition to color, height, and location. To improve the performance of target identification in a dynamic environment, an online boosting method was adapted by continuously updating the features in every frame. Through extensive real-life experiments, the effectiveness of the proposed method was demonstrated by showing experimental results that it outperformed the previous methods.

Pathological gait clustering in post-stroke patients using motion capture data.

BACKGROUND: Analyzing the complex gait patterns of post-stroke patients with lower limb paralysis is essential for rehabilitation. RESEARCH QUESTION: Is it feasible to use the full joint-level kinematic features extracted from the motion capture data of patients directly to identify the optimal gait types that ensure high classification performance? METHODS: In this study, kinematic features were extracted from 111 gait cycle data on joint angles, and angular velocities of 36 post-stroke patients were collected eight times over six months using a motion capture system. Simultaneous clustering and classification were applied to determine the optimal gait types for reliable classification performance. RESULTS: In the given dataset, six optimal gait groups were identified, and the clustering and classification performances were denoted by a silhouette coefficient of 0.1447 and F1 score of 1.0000, respectively. SIGNIFICANCE: There is no distinct clinical classification of post-stroke hemiplegic gaits. However, in contrast to previous studies, more optimal gait types with a high classification performance fully utilizing the kinematic features were identified in this study.

A Novel Robot-Assisted Kinematic Measure for Children with Attention-Deficit/Hyperactivity Disorder: A Preliminary Study.

OBJECTIVE: Attention-deficit/hyperactivity disorder (ADHD) is a neurodevelopmental disorder characterized by inattention, hyperactivity, and impulsivity. In contrast to neurocognitive measurements of inattention and impulsivity, there has been limited research regarding the objective measurement of hyperactivity in youths with ADHD. The purpose of the present study was to investigate the clinical effectiveness of a newly developed Robot-assisted Kinematic Measure for ADHD (RAKMA) in children with ADHD. METHODS: In total, 35 children with ADHD aged 5 to 12 years and 50 healthy controls (HCs) were recruited, and the parents completed the Child Behavior Checklist and the Korean ADHD Diagnostic Scale. RAKMA performance was represented by RAKMA stimulus-response and hyperactivity variables. We compared the RAKMA performance of those with ADHD and with that of HCs and also investigated the correlation between the RAKMA variables and ADHD clinical scale scores. RESULTS: Significant differences between the ADHD and HC groups were observed regarding most RAKMA variables, including correct reactions, commission errors, omission errors, reaction times, migration distance, and migration speed scores. Significant correlations were detected between various ADHD clinical scale scores and RAKMA variables. CONCLUSION: The RAKMA was a clinically useful tool for objectively measuring hyperactivity symptoms in children with ADHD. Further studies with larger samples are warranted.

Analysis of Health Insurance Big Data for Early Detection of Disabilities: Algorithm Development and Validation.

BACKGROUND: Early detection of childhood developmental delays is very important for the treatment of disabilities. OBJECTIVE: To investigate the possibility of detecting childhood developmental delays leading to disabilities before clinical registration by analyzing big data from a health insurance database. METHODS: In this study, the data from children, individuals aged up to 13 years (n=2412), from the Sample Cohort 2.0 DB of the Korea National Health Insurance Service were organized by age range. Using 6 categories (having no disability, having a physical disability, having a brain lesion, having a visual impairment, having a hearing impairment, and having other conditions), features were selected in the order of importance with a tree-based model. We used multiple classification algorithms to find the best model for each age range. The earliest age range with clinically significant performance showed the age at which conditions can be detected early. RESULTS: The disability detection model showed that it was possible to detect disabilities with significant accuracy even at the age of 4 years, about a year earlier than the mean diagnostic age of 4.99 years. CONCLUSIONS: Using big data analysis, we discovered the possibility of detecting disabilities earlier than clinical diagnoses, which would allow us to take appropriate action to prevent disabilities.

Deep-Learning-Based Indoor Human Following of Mobile Robot Using Color Feature.

Human following is one of the fundamental functions in human-robot interaction for mobile robots. This paper shows a novel framework with state-machine control in which the robot tracks the target person in occlusion and illumination changes, as well as navigates with obstacle avoidance while following the target to the destination. People are detected and tracked using a deep learning algorithm, called Single Shot MultiBox Detector, and the target person is identified by extracting the color feature using the hue-saturation-value histogram. The robot follows the target safely to the destination using a simultaneous localization and mapping algorithm with the LIDAR sensor for obstacle avoidance. We performed intensive experiments on our human following approach in an indoor environment with multiple people and moderate illumination changes. Experimental results indicated that the robot followed the target well to the destination, showing the effectiveness and practicability of our proposed system in the given environment.

A fully integrated computer-aided diagnosis system for digital X-ray mammograms via deep learning detection, segmentation, and classification.

A computer-aided diagnosis (CAD) system requires detection, segmentation, and classification in one framework to assist radiologists efficiently in an accurate diagnosis. In this paper, a completely integrated CAD system is proposed to screen digital X-ray mammograms involving detection, segmentation, and classification of breast masses via deep learning methodologies. In this work, to detect breast mass from entire mammograms, You-Only-Look-Once (YOLO), a regional deep learning approach, is used. To segment the mass, full resolution convolutional network (FrCN), a new deep network model, is proposed and utilized. Finally, a deep convolutional neural network (CNN) is used to recognize the mass and classify it as either benign or malignant. To evaluate the proposed integrated CAD system in terms of the accuracies of detection, segmentation, and classification, the publicly available and annotated INbreast database was utilized. The evaluation results of the proposed CAD system via four-fold cross-validation tests show that a mass detection accuracy of 98.96%, Matthews correlation coefficient (MCC) of 97.62%, and F1-score of 99.24% are achieved with the INbreast dataset. Moreover, the mass segmentation results via FrCN produced an overall accuracy of 92.97%, MCC of 85.93%, and Dice (F1-score) of 92.69% and Jaccard similarity coefficient metrics of 86.37%, respectively. The detected and segmented masses were classified via CNN and achieved an overall accuracy of 95.64%, AUC of 94.78%, MCC of 89.91%, and F1-score of 96.84%, respectively. Our results demonstrate that the proposed CAD system, through all stages of detection, segmentation, and classification, outperforms the latest conventional deep learning methodologies. Our proposed CAD system could be used to assist radiologists in all stages of detection, segmentation, and classification of breast masses.

Skin lesion segmentation in dermoscopy images via deep full resolution convolutional networks.

BACKGROUND AND OBJECTIVE: Automatic segmentation of skin lesions in dermoscopy images is still a challenging task due to the large shape variations and indistinct boundaries of the lesions. Accurate segmentation of skin lesions is a key prerequisite step for any computer-aided diagnostic system to recognize skin melanoma. METHODS: In this paper, we propose a novel segmentation methodology via full resolution convolutional networks (FrCN). The proposed FrCN method directly learns the full resolution features of each individual pixel of the input data without the need for pre- or post-processing operations such as artifact removal, low contrast adjustment, or further enhancement of the segmented skin lesion boundaries. We evaluated the proposed method using two publicly available databases, the IEEE International Symposium on Biomedical Imaging (ISBI) 2017 Challenge and PH2 datasets. To evaluate the proposed method, we compared the segmentation performance with the latest deep learning segmentation approaches such as the fully convolutional network (FCN), U-Net, and SegNet. RESULTS: Our results showed that the proposed FrCN method segmented the skin lesions with an average Jaccard index of 77.11% and an overall segmentation accuracy of 94.03% for the ISBI 2017 test dataset and 84.79% and 95.08%, respectively, for the PH2 dataset. In comparison to FCN, U-Net, and SegNet, the proposed FrCN outperformed them by 4.94%, 15.47%, and 7.48% for the Jaccard index and 1.31%, 3.89%, and 2.27% for the segmentation accuracy, respectively. Furthermore, the proposed FrCN achieved a segmentation accuracy of 95.62% for some representative clinical benign cases, 90.78% for the melanoma cases, and 91.29% for the seborrheic keratosis cases in the ISBI 2017 test dataset, exhibiting better performance than those of FCN, U-Net, and SegNet. CONCLUSIONS: We conclude that using the full spatial resolutions of the input image could enable to learn better specific and prominent features, leading to an improvement in the segmentation performance.

Simultaneous detection and classification of breast masses in digital mammograms via a deep learning YOLO-based CAD system.

BACKGROUND AND OBJECTIVE: Automatic detection and classification of the masses in mammograms are still a big challenge and play a crucial role to assist radiologists for accurate diagnosis. In this paper, we propose a novel Computer-Aided Diagnosis (CAD) system based on one of the regional deep learning techniques, a ROI-based Convolutional Neural Network (CNN) which is called You Only Look Once (YOLO). Although most previous studies only deal with classification of masses, our proposed YOLO-based CAD system can handle detection and classification simultaneously in one framework. METHODS: The proposed CAD system contains four main stages: preprocessing of mammograms, feature extraction utilizing deep convolutional networks, mass detection with confidence, and finally mass classification using Fully Connected Neural Networks (FC-NNs). In this study, we utilized original 600 mammograms from Digital Database for Screening Mammography (DDSM) and their augmented mammograms of 2,400 with the information of the masses and their types in training and testing our CAD. The trained YOLO-based CAD system detects the masses and then classifies their types into benign or malignant. RESULTS: Our results with five-fold cross validation tests show that the proposed CAD system detects the mass location with an overall accuracy of 99.7%. The system also distinguishes between benign and malignant lesions with an overall accuracy of 97%. CONCLUSIONS: Our proposed system even works on some challenging breast cancer cases where the masses exist over the pectoral muscles or dense regions.

Analysis of Pelvis-Thorax Coordination Patterns of Professional and Amateur Golfers during Golf Swing.

The aim of this research was to quantify the coordination pattern between thorax and pelvis during a golf swing. The coordination patterns were calculated using vector coding technique, which had been applied to quantify the coordination changes in coupling angle (γ) between two different segments. For this, fifteen professional and fifteen amateur golfers who had no significant history of musculoskeletal injuries. There was no significant difference in coordination patterns between the two groups for rotation motion during backswing (p = 0.333). On the other hand, during the downswing phase, there were significant differences between professional and amateur groups in all motions (flexion/extension: professional [γ] = 187.8°, amateur [γ] = 167.4°; side bending: professional [γ] = 288.4°, amateur [γ] = 245.7°; rotation: professional [γ] = 232.0°, amateur [γ] = 229.5°). These results are expected to be a discriminating measure to assess complex coordination of golfers' trunk movements and preliminary study for interesting comparison by golf skilled levels.

Inter-joint coordination between hips and trunk during downswings: Effects on the clubhead speed.

Understanding of the inter-joint coordination between rotational movement of each hip and trunk in golf would provide basic knowledge regarding how the neuromuscular system organises the related joints to perform a successful swing motion. In this study, we evaluated the inter-joint coordination characteristics between rotational movement of the hips and trunk during golf downswings. Twenty-one right-handed male professional golfers were recruited for this study. Infrared cameras were installed to capture the swing motion. The axial rotation angle, angular velocity and inter-joint coordination were calculated by the Euler angle, numerical difference method and continuous relative phase, respectively. A more typical inter-joint coordination demonstrated in the leading hip/trunk than trailing hip/trunk. Three coordination characteristics of the leading hip/trunk reported a significant relationship with clubhead speed at impact (r < -0.5) in male professional golfers. The increased rotation difference between the leading hip and trunk in the overall downswing phase as well as the faster rotation of the leading hip compared to that of the trunk in the early downswing play important roles in increasing clubhead speed. These novel inter-joint coordination strategies have the great potential to use a biomechanical guideline to improve the golf swing performance of unskilled golfers.

Structural brain changes after traditional and robot-assisted multi-domain cognitive training in community-dwelling healthy elderly.

UNLABELLED: The purpose of this study was to investigate if multi-domain cognitive training, especially robot-assisted training, alters cortical thickness in the brains of elderly participants. A controlled trial was conducted with 85 volunteers without cognitive impairment who were 60 years old or older. Participants were first randomized into two groups. One group consisted of 48 participants who would receive cognitive training and 37 who would not receive training. The cognitive training group was randomly divided into two groups, 24 who received traditional cognitive training and 24 who received robot-assisted cognitive training. The training for both groups consisted of daily 90-min-session, five days a week for a total of 12 weeks. The primary outcome was the changes in cortical thickness. When compared to the control group, both groups who underwent cognitive training demonstrated attenuation of age related cortical thinning in the frontotemporal association cortices. When the robot and the traditional interventions were directly compared, the robot group showed less cortical thinning in the anterior cingulate cortices. Our results suggest that cognitive training can mitigate age-associated structural brain changes in the elderly. TRIAL REGISTRATION: ClnicalTrials.gov NCT01596205.

Robust extraction of P300 using constrained ICA for BCI applications.

P300 is a positive event-related potential used by P300-brain computer interfaces (BCIs) as a means of communication with external devices. One of the main requirements of any P300-based BCI is accuracy and time efficiency for P300 extraction and detection. Among many attempted techniques, independent component analysis (ICA) is currently the most popular P300 extraction technique. However, since ICA extracts multiple independent components (ICs), its use requires careful selection of ICs containing P300 responses, which limits the number of channels available for computational efficiency. Here, we propose a novel procedure for P300 extraction and detection using constrained independent component analysis (cICA) through which we can directly extract only P300-relevant ICs. We tested our procedure on two standard datasets collected from healthy and disabled subjects. We tested our procedure on these datasets and compared their respective performances with a conventional ICA-based procedure. Our results demonstrate that the cICA-based method was more reliable and less computationally expensive, and was able to achieve 97 and 91.6% accuracy in P300 detection from healthy and disabled subjects, respectively. In recognizing target characters and images, our approach achieved 95 and 90.25% success in healthy and disabled individuals, whereas use of ICA only achieved 83 and 72.25%, respectively. In terms of information transfer rate, our results indicate that the ICA-based procedure optimally performs with a limited number of channels (typically three), but with a higher number of available channels (>3), its performance deteriorates and the cICA-based one performs better.