Explaining Machine Learning Models for Clinical Gait Analysis

Machine Learning (ML) is increasingly used to support decision-making in the healthcare sector. While ML approaches provide promising results with regard to their classification performance, most share a central limitation, their black-box character. This article investigates the usefulness of Explainable Artificial Intelligence (XAI) methods to increase transparency in automated clinical gait classification based on time series. For this purpose, predictions of state-of-the-art classification methods are explained with a XAI method called Layer-wise Relevance Propagation (LRP). Our main contribution is an approach that explains class-specific characteristics learned by ML models that are trained for gait classification. We investigate several gait classification tasks and employ different classification methods, i.e., Convolutional Neural Network, Support Vector Machine, and Multi-layer Perceptron. We propose to evaluate the obtained explanations with two complementary approaches: a statistical analysis of the underlying data using Statistical Parametric Mapping and a qualitative evaluation by two clinical experts. A gait dataset comprising ground reaction force measurements from 132 patients with different lower-body gait disorders and 62 healthy controls is utilized. Our experiments show that explanations obtained by LRP exhibit promising statistical properties concerning inter-class discriminativity and are also in line with clinically relevant biomechanical gait characteristics.

Deep Learning-Aided Automatic Contouring of Clinical Target Volumes for Radiotherapy in Breast Cancer After Modified Radical Mastectomy

Purpose: The aim of this study is to develop a practicable automatic clinical target volume (CTV) delineation method for radiotherapy of breast cancer after modified radical mastectomy.Methods: Unlike breast conserving surgery, the radiotherapy CTV for modified radical mastectomy involves several regions, including CTV in the chest wall (CTVcw), supra- and infra-clavicular region (CTVsc), and internal mammary lymphatic region (CTVim). For accurate and efficient segmentation of the CTVs in radiotherapy of breast cancer after modified radical mastectomy, a multi-scale convolutional neural network with an orientation attention mechanism is proposed to capture the corresponding features in different perception fields. A channel-specific local Dice loss, alongside several data augmentation methods, is also designed specifically to stabilize the model training and improve the generalization performance of the model. The segmentation performance is quantitatively evaluated by statistical metrics and qualitatively evaluated by clinicians in terms of consistency and time efficiency.Results: The proposed method is trained and evaluated on the self-collected dataset, which contains 110 computed tomography scans from patients with breast cancer who underwent modified mastectomy. The experimental results show that the proposed segmentation method achieved superior performance in terms of Dice similarity coefficient (DSC), Hausdorff distance (HD) and Average symmetric surface distance (ASSD) compared with baseline approaches.Conclusion: Both quantitative and qualitative evaluation results demonstrated that the specifically designed method is practical and effective in automatic contouring of CTVs for radiotherapy of breast cancer after modified radical mastectomy. Clinicians can significantly save time on manual delineation while obtaining contouring results with high consistency by employing this method.

Enhancing engagement in meaningful occupation in a dementia town: A qualitative evaluation of MOED – The meaningful occupational engagement intervention for people with dementia

Introduction Engagement in meaningful everyday life occupations is linked to well-being. However, people with dementia are often deprived of engagement. As a response, a Danish Dementia Town was established with the intention of transforming care services to improve opportunities for meaningful engagement. The evidence-based The meaningful occupational engagement intervention for people with dementia (MOED) intervention was developed and implemented in dementia town to enhance meaningful occupational engagement. The purpose of this paper is to evaluate the content, impact and implementation process of MOED. Method To evaluate MOED, we applied a program theory-based qualitative approach, building on participant observations and informal conversations with people with dementia ( n = 7) and staff ( n = 9). Data were analysed from a hermeneutic approach to gain an in-depth understanding of how MOED influenced meaningful occupational engagement and to evaluate the implementation process. Findings The main theme ‘Engagement in meaningful occupations – a conditional and fragile process’ emerged along with three subthemes: ‘Creating an everyday space of meaning: Conditions of the intervention’, ‘Occupational engagement as an opportunity to blossom: Impact of the intervention’ and ‘Professional identity, culture and reflections: Contextual barriers to changes to working practice’. Together, the subthemes illustrate how engagement in meaningful occupations arose. However, opportunities to engage in meaningful occupations were fragile, as they depended on various factors within the context. Conclusion Engagement in meaningful occupations emerged when MOED was applied in accordance with the program theory, as it seemed to create spaces where people with dementia could engage in meaningful occupations and they could blossom over time. However, MOED was only partially implemented in accordance with the developed program theory, as several contextual barriers influenced the sustainability of the intervention. MOED showed potential to support improvement in dementia care activity programmes to enhance opportunities to engage in meaningful occupations in everyday life for people living with dementia.

Data Checkers: A Grid-Based UI for Managing Patient-Generated Data Sharing to Support Collaborative Self-Care

Chronic health conditions are becoming increasingly prevalent. As part of chronic care, sharing patient-generated health data (PGHD) is likely to play a prominent role. Sharing PGHD is increasingly recognized as potentially useful for not only monitoring health conditions but for informing and supporting collaboration with caregivers and healthcare providers. In this paper, we describe a new design for the fine-grained control over sharing one's PGHD to support collaborative self-care, one that centers on giving people with health conditions control over their own data. The system, Data Checkers (DC), uses a grid-based interface and a preview feature to provide users with the ability to control data access and dissemination. DC is of particular use in the case of severe chronic conditions, such as spinal cord injuries and disorders (SCI/D), that require not just intermittent involvement of healthcare providers but daily support and assistance from caregivers. In this paper, after providing relevant background information, we articulate our steps for developing this innovative system for sharing PGHD including (a) use of a co-design process; (b) identification of design requirements; and (c) creation of the DC System. We then present a qualitative evaluation of DC to show how DC satisfied these design requirements in a way that provided advantages for care. Our work extends existing research in the areas of Human-Computer Interaction (HCI), Computer-Supported Cooperative Work (CSCW), Ubiquitous Computing (Ubicomp), and Health Informatics about sharing data and PGHD.

Multi-inquiry hybrid e-learning: Instructor-learner experiences

Background: Hybrid teaching methodologies involve the purposeful combination of traditional teaching with technology advances. Despite some challenges, they have gained popularity recently, especially during the coronavirus pandemic. This study evaluated hybrid e-learning with multiple inquiries involving students’ receptiveness, preferences, behaviours and instructor observations. Method: The methodology involved a mixed-method approach with a qualitative observational case study, surveys and interviews for problem-based learning alternatives to traditional lectures. Instruction included: 1) Assigned primary literature reading with study questions to be completed before class; 2) Out-of-class online video-clips with visual, practical application (i.e. lithium and non-lithium induced tremor assessment) and online discussion in CANVAS Learning Management System; 3) Start-of-class quiz in ExamSoft, in-class team-based application questions with instructor-led discussion; 4) Out-of-class team final exam review assignment in CANVAS. Results: Qualitative themes were student engagement, flexibility, preferences, academic and non-academic stressors, etiquette, and defining responsibility for academic success. The majority of students preferred primary literature review, video clips, followed by online CANVAS discussions. Written assignments were the least desirable. Conclusion: These experiences are useful for qualitative evaluation of teaching and learning methods.

Development and effectiveness of greenhouse type solar dryer for coriander leaves

Aim: Sun drying system is not able to provide the best drying performance and quality dried produce of leafy vegetables. To facilitate better options to the farmers, this study aims to develop and evaluate a cost-effective greenhouse type solar dryer to improve shelf stability of coriander leaves. Methodology: A greenhouse type solar dryer (5m x 3m x 2.3m) was developed with the provision of rigid frame, 200 µ UV stabilized LDPE glazing material, solar collector cum drying chamber, inlet air and outlet air openings. The dryer was evaluated for its performance, drying characteristics of coriander leaves and qualitative evaluation of dried product as compared to open sun drying. Results: The greenhouse type solar dryer performed well for coriander drying with increased level of temperature inside the dryer (42oC) and 24% reduced drying time as compared to sun drying (29oC).The loading capacity, efficiency, payback period and B:C ratio of the dryer were found to be 100 kg, 70.47 %, 1.26 yr and 1.79, respectively. The drying of coriander leaves occured mostly in falling rate period. Coriander leaves dried under the dryer possessed higher values of chlorophyll content (1.356 mg g-1 of tissue), ascorbic acid content (111.257 mg 100g-1), rehydration ratio (5.302) and coefficient of rehydration (0.762) than sun dried coriander leaves with the respective values of these quality parameters being 1.097 mg g-1 of tissue, 62.37 mg 100g-1, 4.715 and 0.689, respectively.

The Effect of Deoxyfluorination on Intermolecular Interactions in the Crystal Structures of 1,6-Anhydro-2,3-epimino-hexopyranoses

The effect of substitution on intermolecular interactions was investigated in a series of 1,6-anhydro-2,3-epimino-hexopyranoses. The study focused on the qualitative evaluation of intermolecular interactions using DFT calculations and the comparison of molecular arrangements in the crystal lattice. Altogether, ten crystal structures were compared, including two structures of C4-deoxygenated, four C4-deoxyfluorinated and four parent epimino pyranoses. It was found that the substitution of the original hydroxy group by hydrogen or fluorine leads to a weakening of the intermolecular interaction by approximately 4 kcal/mol. The strength of the intermolecular interactions was found to be in the following descending order: hydrogen bonding of hydroxy groups, hydrogen bonding of the amino group, interactions with fluorine and weak electrostatic interactions. The intermolecular interactions that involved fluorine atom were rather weak; however, they were often supported by other weak interactions. The fluorine atom was not able to substitute the role of the hydroxy group in molecular packing and the fluorine atoms interacted only weakly with the hydrogen atoms located at electropositive regions of the carbohydrate molecules. However, the fluorine interaction was not restricted to a single molecule but was spread over at least three other molecules. This feature is a base for similar molecule arrangements in the structures of related compounds, as we found for the C4-Fax and C4-Feq epimines presented here.