Dimensional emotion recognition from camera-based PRV features.

Heart rate variability (HRV) is an important indicator of autonomic nervous system activity and can be used for the identification of affective states. The development of remote Photoplethysmography (rPPG) technology has made it possible to measure pulse rate variability (PRV) using a camera without any sensor-skin contact, which is highly correlated to HRV, thus, enabling contactless assessment of emotional states. In this study, we employed ten machine learning techniques to identify emotions using camera-based PRV features. Our experimental results show that the best classification model achieved a coordination correlation coefficient of 0.34 for value recognition and 0.36 for arousal recognition. The rPPG-based measurement has demonstrated promising results in detecting HAHV (high-arousal high-valence) emotions with high accuracy. Furthermore, for emotions with less noticeable variations, such as sadness, the rPPG-based measure outperformed the baseline deep network for facial expression analysis.

Model-Driven Dementia Prevention and Intervention Platform.

Most types of dementia, including Alzheimer's disease, are not curable. However, there are risk factors, such as obesity or hypertension, that can promote the development of dementia. Holistic treatment of these risk factors can prevent the onset of dementia or delay it in its early stages. To support individualized treatment of risk factors in dementia, this paper presents a model-driven digital platform. It enables monitoring of biomarkers using smart devices from the internet of medical things (IoMT) for the target group. The collected data from such devices can be used to optimize and adjust treatment in a patient in the loop manner. To this end, providers such as Google Fit and Withings have been connected to the platform as example data sources. To achieve treatment and monitoring data interoperability with existing medical systems, internationally accepted standards such as FHIR are used. The configuration and control of the personalized treatment processes are achieved using a self-developed domain-specific language. For this language, an associated diagram editor was implemented, which allows the management of the treatment processes through graphical models. This graphical representation should help treatment providers to understand and manage these processes more easily. To investigate this hypothesis, a usability study was conducted with twelve participants. We were able to show that such graphical representations provide advantages in clarity in reviewing the system, but lack in easy set-up (compared to wizard-style systems).

Digital Health Apps in the Context of Dementia: Questionnaire Study to Assess the Likelihood of Use Among Physicians.

BACKGROUND: Age-related diseases such as dementia are playing an increasingly important role in global population development. Thus, prevention, diagnostics, and interventions require more accessibility, which can be realized through digital health apps. With the app on prescription, Germany made history by being the first country worldwide to offer physicians the possibility to prescribe and reimburse digital health apps as of the end of the year 2020. OBJECTIVE: Considering the lack of knowledge about correlations with the likelihood of use among physicians, this study aimed to address the question of what makes the use of a digital health app by physicians more likely. METHODS: We developed and validated a novel measurement tool-the Digital Health Compliance Questionnaire (DHCQ)-in an interdisciplinary collaboration of experts to assess the role of proposed factors in the likelihood of using a health app. Therefore, a web-based survey was conducted to evaluate the likelihood of using a digital app called DemPredict to screen for Alzheimer dementia. Within this survey, 5 latent dimensions (acceptance, attitude toward technology, technology experience, payment for time of use, and effort of collection), the dependent variable likelihood of use, and answers to exploratory questions were recorded and tested within directed correlations. Following a non-probability-sampling strategy, the study was completed by 331 physicians from Germany in the German language, of whom 301 (90.9%) fulfilled the study criteria (eg, being in regular contact with patients with dementia). These data were analyzed using a range of statistical methods to validate the dimensions of the DHCQ. RESULTS: The DHCQ revealed good test theoretical measures-it showed excellent fit indexes (Tucker-Lewis index=0.98; comparative fit index=0.982; standardized root mean square residual=0.073; root mean square error of approximation=0.037), good internal consistency (Cronbach α=.83), and signs of moderate to large correlations between the DHCQ dimensions and the dependent variable. The correlations between the variables acceptance, attitude toward technology, technology experience, and payment for the time of use and the dependent variable likelihood of use ranged from 0.29 to 0.79, and the correlation between effort of the collection and likelihood of use was -0.80. In addition, we found high levels of skepticism regarding data protection, and the age of the participants was found to be negatively related to their technical experience and attitude toward technology. CONCLUSIONS: In the context of the results, increased communication between the medical and technology sectors and significantly more awareness raising are recommended to make the use of digital health apps more attractive to physicians as they can be adjusted to their everyday needs. Further research could explore the connection between areas such as adherence on the patient side and its impact on the likelihood of use by physicians.

Explainable Artificial Intelligence in Ambulatory Digital Dementia Screenings.

Recently, digital apps have entered the market to enable the early diagnosis of dementia by offering digital dementia screenings. Some of these apps use Machine Learning (ML) to predict cognitive impairment. The aim of this work is to find explanations for the predictions of such a mobile application called DemPredict using methods from the field of Explainable Artificial Intelligence (XAI). In order to evaluate which method is best suited, different XAI approaches are used and compared. However, the comparability of the results is a key challenge. By evaluating the trustworthiness, stability, and computation time of the methods, it is possible to identify the optimal XAI approaches for the respective algorithms.