Background: Artificial intelligence (AI) is giving rise to a revolution in medicine and health care. Mental health conditions are highly prevalent in many countries, and the COVID-19 pandemic has increased the risk of further erosion of the mental well-being in the population. Therefore, it is relevant to assess the current status of the application of AI toward mental health research to inform about trends, gaps, opportunities, and challenges. Objective: This study aims to perform a systematic overview of AI applications in mental health in terms of methodologies, data, outcomes, performance, and quality. Methods: A systematic search in PubMed, Scopus, IEEE Xplore, and Cochrane databases was conducted to collect records of use cases of AI for mental health disorder studies from January 2016 to November 2021. Records were screened for eligibility if they were a practical implementation of AI in clinical trials involving mental health conditions. Records of AI study cases were evaluated and categorized by the International Classification of Diseases 11th Revision (ICD-11). Data related to trial settings, collection methodology, features, outcomes, and model development and evaluation were extracted following the CHARMS (Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies) guideline. Further, evaluation of risk of bias is provided. Results: A total of 429 nonduplicated records were retrieved from the databases and 129 were included for a full assessment—18 of which were manually added. The distribution of AI applications in mental health was found unbalanced between ICD-11 mental health categories. Predominant categories were Depressive disorders (n=70) and Schizophrenia or other primary psychotic disorders (n=26). Most interventions were based on randomized controlled trials (n=62), followed by prospective cohorts (n=24) among observational studies. AI was typically applied to evaluate quality of treatments (n=44) or stratify patients into subgroups and clusters (n=31). Models usually applied a combination of questionnaires and scales to assess symptom severity using electronic health records (n=49) as well as medical images (n=33). Quality assessment revealed important flaws in the process of AI application and data preprocessing pipelines. One-third of the studies (n=56) did not report any preprocessing or data preparation. One-fifth of the models were developed by comparing several methods (n=35) without assessing their suitability in advance and a small proportion reported external validation (n=21). Only 1 paper reported a second assessment of a previous AI model. Risk of bias and transparent reporting yielded low scores due to a poor reporting of the strategy for adjusting hyperparameters, coefficients, and the explainability of the models. International collaboration was anecdotal (n=17) and data and developed models mostly remained private (n=126). Conclusions: These significant shortcomings, alongside the lack of information to ensure reproducibility and transparency, are indicative of the challenges that AI in mental health needs to face before contributing to a solid base for knowledge generation and for being a support tool in mental health management.

To describe the development, as part of the European Union MOSAIC (Models and Simulation Techniques for Discovering Diabetes Influence Factors) project, of a dashboard-based system for the management of type 2 diabetes and assess its impact on clinical practice. The MOSAIC dashboard system is based on predictive modeling, longitudinal data analytics, and the reuse and integration of data from hospitals and public health repositories. Data are merged into an i2b2 data warehouse, which feeds a set of advanced temporal analytic models, including temporal abstractions, care-flow mining, drug exposure pattern detection, and risk-prediction models for type 2 diabetes complications. The dashboard has 2 components, designed for (1) clinical decision support during follow-up consultations and (2) outcome assessment on populations of interest. To assess the impact of the clinical decision support component, a pre-post study was conducted considering visit duration, number of screening examinations, and lifestyle interventions. A pilot sample of 700 Italian patients was investigated. Judgments on the outcome assessment component were obtained via focus groups with clinicians and health care managers. The use of the decision support component in clinical activities produced a reduction in visit duration (P ≪ .01) and an increase in the number of screening exams for complications (P < .01). We also observed a relevant, although nonstatistically significant, increase in the proportion of patients receiving lifestyle interventions (from 69% to 77%). Regarding the outcome assessment component, focus groups highlighted the system’s capability of identifying and understanding the characteristics of patient subgroups treated at the center. Our study demonstrates that decision support tools based on the integration of multiple-source data and visual and predictive analytics do improve the management of a chronic disease such as type 2 diabetes by enacting a successful implementation of the learning health care system cycle.

Health and sociological indicators alert that life expectancy is increasing, hence so are the years that patients have to live with chronic diseases and co-morbidities. With the advancement in ICT, new tools and paradigms are been explored to provide effective and efficient health care. Telemedicine and health sensors stand as indispensable tools for promoting patient engagement, self-management of diseases and assist doctors to remotely follow up patients. In this paper, we evaluate a rapid prototyping solution for information merging based on five health sensors and two low-cost ubiquitous computing components: Arduino and Raspberry Pi. Our study, which is entirely described with the purpose of reproducibility, aimed to evaluate the extent to which portable technologies are capable of integrating wearable sensors by comparing two deployment scenarios: Raspberry Pi 3 and Personal Computer. The integration is implemented using a choreography engine to transmit data from sensors to a display unit using web services and a simple communication protocol with two modes of data retrieval. Performance of the two set-ups is compared by means of the latency in the wearable data transmission and data loss. PC has a delay of 0.051 ± 0.0035 s (max = 0.2504 s), whereas the Raspberry Pi yields a delay of 0.0175 ± 0.149 s (max = 0.294 s) for N = 300. Our analysis confirms that portable devices ( $p<<0.01$ ) are suitable to support the transmission and analysis of biometric signals into scalable telemedicine systems.

Expressing anthropometric parameters (height, weight, BMI) as z-score is a key principle in the clinical assessment of children and adolescents. The Centre for Disease Control and Prevention (CDC) growth charts and the CDC-LMS method for z-score calculation are widely used to assess growth and nutritional status, though they can be imprecise in some percentiles. To improve the accuracy of z-score calculation by revising the statistical method using the original data used to develop current z-score calculators. A Gaussian Process Regressions (GPR) was designed and internally validated. Z-scores for weight-for-age (WFA), height-for-age (HFA) and BMI-for-age (BMIFA) were compared with WHO and CDC-LMS methods in 1) standard z-score cut-off points, 2) simulated population of 3000 children and 3) real observations 212 children aged 2 to 18 yo. GPR yielded more accurate calculation of z-scores for standard cut-off points (p<<0.001) with respect to CDC-LMS and WHO approaches. WFA, HFA and BMIFA z-score calculations based on the 3 different methods using simulated and real patients, showed a large variation irrespective of gender and age. Z-scores around 0 +/- 1 showed larger variation than the values above and below +/- 2. The revised z-score calculation method was more accurate than CDC-LMS and WHO methods for standard cut-off points. On simulated and real data, GPR based calculation provides more accurate z-score determinations, and thus, a better classification of patients below and above cut-off points. Statisticians and clinicians should consider the potential benefits of updating their calculation method for an accurate z-score determination.

The application of Value-based Healthcare requires not only the identification of key processes in the clinical domain but also an adequate analysis of the value chain delivered to the patient. Data Science and Big Data approaches are technologies that enable the creation of accurate systems that model reality. However, classical Data Mining techniques are presented by professionals as black boxes. This evokes a lack of trust in those techniques in the medical domain. Process Mining technologies are human-understandable Data Science tools that can fill this gap to support the application of Value-Based Healthcare in real domains. The aim of this paper is to perform an analysis of the ways in which Process Mining techniques can support health professionals in the application of Value-Based Technologies. For this purpose, we explored these techniques by analyzing emergency processes and applying the critical timing of Stroke treatment and a Question-Driven methodology. To demonstrate the possibilities of Process Mining in the characterization of the emergency process, we used a real log with 9046 emergency episodes from 2145 stroke patients that occurred from January 2010 to June 2017. Our results demonstrate how Process Mining technology can highlight the differences between the flow of stroke patients compared with that of other patients in an emergency. Further, we show that support for health professionals can be provided by improving their understanding of these techniques and enhancing the quality of care.

Research in type 1 diabetes management has increased exponentially since the irruption of mobile health apps for its remote and self-management. Despite this fact, the features affect in the disease management and patient empowerment are adopted by app makers and provided to the general population remain unexplored. To study the gap between literature and available apps for type 1 diabetes self-management and patient empowerment and to discover the features that an ideal app should provide to people with diabetes. The methodology comprises systematic reviews in the scientific literature and app marketplaces. We included articles describing interventions that demonstrated an effect on diabetes management with particular clinical endpoints through the use of mobile technologies. The features of these apps were gathered in a taxonomy of what an ideal app should look like to then assess which of these features are available in the market. The literature search resulted in 231 matches. Of these, 55 met the inclusion criteria. A taxonomy featuring 3 levels of characteristics was designed based on 5 papers which were selected for the synthesis. Level 1 includes 10 general features (Personalization, Family support, Agenda, Data record, Insulin bolus calculator, Data management, Interaction, Tips and support, Reminders, and Rewards) Level 2 and Level 3 included features providing a descriptive detail of Level 1 features. Eighty apps matching the inclusion criteria were analyzed. None of the assessed apps fulfilled the features of the taxonomy of an ideal app. Personalization (70/80, 87.5%) and Data record (64/80, 80.0%) were the 2 top prevalent features, whereas Agenda (5/80, 6.3%) and Rewards (3/80, 3.8%) where the less predominant. The operating system was not associated with the number of features (P=.42, F=.81) nor the type of feature (P=.20, χ2=11.7). Apps were classified according to the number of level 1 features and sorted into quartiles. First quartile apps had a regular distribution of the ten features in the taxonomy whereas the other 3 quartiles had an irregular distribution. There are significant gaps between research and the market in mobile health for type 1 diabetes management. While the literature focuses on aspects related to gamification, rewarding, and social communities, the available apps are focused on disease management aspects such as data record and appointments. Personalized and tailored empowerment features should be included in commercial apps for large-scale assessment of potential in the self-management of the disease.