Background: Electronic health records and IT infrastructure in primary care allow for digital documentation and access to information, which can be used to guide evidence-based care and monitor patient safety and quality of care. Quality indicators specified by regulatory authorities can be automatically computed and presented to primary care staff. However, the implementation of digital information systems (DIS) in health care can be challenging, and understanding factors such as relative advantage, compatibility, complexity, trialability, and observability is needed to improve the success and rate of adoption and diffusion.

Objective: This study aims to explore how DIS are used and perceived by health care professionals in primary care.

Methods: This study used quantitative assessment to gather survey data on the use and potential of DIS in health care in Sweden from the perspectives of primary care personnel in various roles. The digital questionnaire was designed to be short and contained 3 sections covering respondent characteristics, current use of platforms, and perceptions of decision support tools. Data were analyzed using descriptive statistics, nonparametric hypothesis testing, ordinal coefficient α, and confirmatory factor analysis.

Results: The study collected responses from participants across 10 regions of Sweden, comprising 31.9% (n=22) from private clinics and 68.1% (n=47) from public clinics. Participants included administrators (18/69, 26.1%), a medical strategist (1/69, 1.4%), and physicians (50/69, 72.5%). Usage frequency varied as follows: 11.6% (n=8) used DIS weekly, 24.6% (n=17) monthly, 27.5% (n=19) a few times a year, 26.1% (n=18) very rarely, and 10.1% (n=7) lacked access. Administrators used DIS more frequently than physicians (P=.005). DIS use centered on quality improvement and identifying high-risk patients, with differences by role. Physicians were more inclined to use DIS out of curiosity (P=.01). Participants desired DIS for patient follow-up, lifestyle guidance, treatment suggestions, reminders, and shared decision-making. Administrators favored predictive analysis (P<.001), while physicians resisted immediate patient identification (P=.03). The 5 innovation attributes showed high internal consistency (α>.7). These factors explained 78.5% of questionnaire variance, relating to complexity, competitive advantage, compatibility, trialability, and observability. Factors 2, 3, and 4 predicted intention to use DIS, with factor 2 alone achieving the best accuracy (root-mean-square=0.513).

Conclusions: Administrators and physicians exhibited role-based DIS use patterns highlighting the need for tailored approaches to promote DIS adoption. The study reveals a link between positive perceptions and intention to use DIS, emphasizing the significance of considering all factors for successful health care integration. The results suggest various directions for future studies. These include refining the trialability and observability questions for increased reliability and validity, investigating a larger sample with more specific target groups to improve generalization, and exploring the relevance of different groups' perspectives and needs in relation to decisions about and use of DIS. ©Anita Sant’Anna, Jens Nygren. 

Activity recognition in smart environments is essential for ensuring the wellbeing of older residents. By tracking activities of daily living (ADLs), a person’s health status can be monitored over time. Nonetheless, accurate activity classification must overcome the fact that each person performs ADLs in different ways and in homes with different layouts. One possible solution is to obtain large amounts of data to train a supervised classifier. Data collection in real environments, however, is very expensive and cannot contain every possible variation of how different ADLs are performed. A more cost-effective solution is to generate a variety of simulated scenarios and synthesize large amounts of data. Nonetheless, simulated data can be considerably different from real data. Therefore, this paper proposes the use of regression models to better approximate real observations based on simulated data. To achieve this, ADL data from a smart home were first compared with equivalent ADLs performed in a simulator. Such comparison was undertaken considering the number of events per activity, number of events per type of sensor per activity, and activity duration. Then, different regression models were assessed for calculating real data based on simulated data. The results evidenced that simulated data can be transformed with a prediction accuracy of R² = 97.03%.

© Springer Science+Business Media, LLC, part of Springer Nature 2020

Background: Information on how behavior change strategies have been used to design digital interventions (DIs) to improve blood pressure (BP) control or medication adherence (MA) for patients with hypertension is currently limited.

Objective: Hypertension is a major modifiable risk factor for cardiovascular diseases and can be controlled with appropriate medication. Many interventions that target MA to improve BP are increasingly using modern digital technologies. This systematic review was conducted to discover how DIs have been designed to improve MA and BP control among patients with hypertension in the recent 10 years. Results were mapped into a matrix of change objectives using the Intervention Mapping framework to guide future development of technologies to improve MA and BP control.

Methods: We included all the studies regarding DI development to improve MA or BP control for patients with hypertension published in PubMed from 2008 to 2018. All the DI components were mapped into a matrix of change objectives using the Intervention Mapping technique by eliciting the key determinant factors (from patient and health care team and system levels) and targeted patient behaviors.

Results: The analysis included 54 eligible studies. The determinants were considered at two levels: patient and health care team and system. The most commonly described determinants at the patient level were lack of education, lack of self-awareness, lack of self-efficacy, and forgetfulness. Clinical inertia and an inadequate health workforce were the most commonly targeted determinants at the health care team and system level. Taking medication, interactive patient-provider communication, self-measurement, and lifestyle management were the most cited patient behaviors at both levels. Most of the DIs did not include support from peers or family members, despite its reported effectiveness and the rate of social media penetration.

Conclusions: This review highlights the need to design a multifaceted DI that can be personalized according to patient behavior(s) that need to be changed to overcome the key determinant(s) of low adherence to medication or uncontrolled BP among patients with hypertension, considering different levels including patient and healthcare team and system involvement. © Kobra Etminani, Arianna Tao Engström, Carina Göransson, Anita Sant’Anna, Sławomir Nowaczyk.

In recent years, there has been increased research interest in detecting anomalies in temporal streaming data. A variety of algorithms have been developed in the data mining community, which can be divided into two categories (i.e., general and ad hoc). In most cases, general approaches assume the one-size-fits-all solution model where a single anomaly detector can detect all anomalies in any domain.  To date, there exists no single general method that has been shown to outperform the others across different anomaly types, use cases and datasets. On the other hand, ad hoc approaches that are designed for a specific application lack flexibility. Adapting an existing algorithm is not straightforward if the specific constraints or requirements for the existing task change. In this paper, we propose SAFARI, a general framework formulated by abstracting and unifying the fundamental tasks in streaming anomaly detection, which provides a flexible and extensible anomaly detection procedure. SAFARI helps to facilitate more elaborate algorithm comparisons by allowing us to isolate the effects of shared and unique characteristics of different algorithms on detection performance. Using SAFARI, we have implemented various anomaly detectors and identified a research gap that motivates us to propose a novel learning strategy in this work. We conducted an extensive evaluation study of 20 detectors that are composed using SAFARI and compared their performances using real-world benchmark datasets with different properties. The results indicate that there is no single superior detector that works well for every case, proving our hypothesis that "there is no free lunch" in the streaming anomaly detection world. Finally, we discuss the benefits and drawbacks of each method in-depth and draw a set of conclusions to guide future users of SAFARI.

Background and purpose: Patients’ adherence to medication is a complex, multidimensional phenomenon. Dispensation data and electronic health records are used to approximate medication-taking through refill adherence. In-depth discussions on the adverse effects of data quality and computational differences are rare. The purpose of this article is to evaluate the impact of common pitfalls when computing medication adherence using electronic health records.

Procedures: We point out common pitfalls associated with the data and operationalization of adherence measures. We provide operational definitions of refill adherence and conduct experiments to determine the effect of the pitfalls on adherence estimations. We performed statistical significance testing on the impact of common pitfalls using a baseline scenario as reference.

Findings: Slight changes in definition can significantly skew refill adherence estimates. Pickup patterns cause significant disagreement between measures and the commonly used proportion of days covered. Common data related issues had a small but statistically significant (p < 0.05) impact on population-level and significant effect on individual cases.

Conclusion: Data-related issues encountered in real-world administrative databases, which affect various operational definitions of refill adherence differently, can significantly skew refill adherence values, leading to false conclusions about adherence, particularly when estimating adherence for individuals. © 2020 The Authors. Published by Elsevier B.V. 

Understanding the heat usage of customers is crucial for effective district heating operations and management. Unfortunately, existing knowledge about customers and their heat load behaviors is quite scarce. Most previous studies are limited to small-scale analyses that are not representative enough to understand the behavior of the overall network. In this work, we propose a data-driven approach that enables large-scale automatic analysis of heat load patterns in district heating networks without requiring prior knowledge. Our method clusters the customer profiles into different groups, extracts their representative patterns, and detects unusual customers whose profiles deviate significantly from the rest of their group. Using our approach, we present the first large-scale, comprehensive analysis of the heat load patterns by conducting a case study on many buildings in six different customer categories connected to two district heating networks in the south of Sweden. The 1222 buildings had a total floor space of 3.4 million square meters and used 1540 TJ heat during 2016. The results show that the proposed method has a high potential to be deployed and used in practice to analyze and understand customers’ heat-use habits. © 2019 Calikus et al. Published by Elsevier Ltd.

Emotions play an important role in human communication, interaction, and decision making processes. Therefore, considerable efforts have been made towards the automatic identification of human emotions, in particular electroencephalogram (EEG) signals and Data Mining (DM) techniques have been then used to create models recognizing the affective states of users. However, most previous works have used clinical grade EEG systems with at least 32 electrodes. These systems are expensive and cumbersome, and therefore unsuitable for usage during normal daily activities. Smaller EEG headsets such as the Emotiv are now available and can be used during daily activities. This paper investigates the accuracy and applicability of previous affective recognition methods on data collected with an Emotiv headset while participants used a personal computer to fulfill several tasks. Several features were extracted from four channels only (AF3, AF4, F3 and F4 in accordance with the 10–20 system). Both Support Vector Machine and Naïve Bayes were used for emotion classification. Results demonstrate that such methods can be used to accurately detect emotions using a small EEG headset during a normal daily activity. © 2018, Springer-Verlag GmbH Germany, part of Springer Nature.

Background: Modern strategies for knee osteoarthritis (OA) treatment and prevention includes early detection and analyses about pain, gait and lower extremity muscle function including both strength and stability. The very first sign of knee OA is pain or perceived knee instability, often experienced during weight bearing activities e.g. walking. Increased muscle strength will provide dynamic joint stability, reduce pain, and disability. Specific measures of gait symmetry (GS) can be assessed objectively by using accelerometers, which potentially is a feasible method when evaluating early symptoms of symptomatic knee OA.

Objectives: The aim was to study if symptoms of early knee pain affected gait symmetry, and the association between lower extremity muscles function and gait symmetry in patients with symptomatic knee OA.

Methods: Thirty-five participants (mean age 52 SD 9 years, 66% women) with uni- or bilateral symptomatic knee OA, and without signs of an inflammatory rheumatic disease or knee trauma were included. Pain was assessed by a numeric rating scale (NRS, range 0-10 best to worse), tests of lower extremity muscle function with the maximum number of one leg rises. Dynamic stability was measured as GS by using wearable inertial sensors (PXNordic senseneering platform), during the 6 min walking test to obtain spatio-temporal gait parameters. GS was computed based on stride time (temporal symmetry, TS) and stride length (spatial symmetry, SS). Stride length was normalized by height. Kruskal-Wallis and Spearman’s correlation coefficient were used for analyses.

Results: Reports of knee pain did not differ between gender (women 4.7, SD 2.4 vs. men 3.9, SD 2.4, p= 0.362), neither did one leg rises or gait symmetry. Participants who reported unilateral knee pain (left/right side n=9/13), had a shorter stride length on the painful side. The mean difference in stride length was 0.7% of the subject’s height (SD 1.3). Participants with unilateral pain also presented less SS gait than those who reported bilateral pain (p=0.005). The higher number of one-leg rises performed, the better TS was observed. We found a significant relationship between TS and one-leg rise for the right r s =-0.39, p=0.006, and left r s =-0.40, p=0.004 left side). No significant relationship was observed between SS and one-leg rises.

Conclusion: Our results is in line with earlier findings stating that knee pain affects GS negatively and that lower extremity muscle function is an important feature for symmetry and dynamic joint stability in patients with symptomatic knee OA. We also found that pain in one leg was related to impaired GS. Bilateral knee pain was however more symmetrical and will need healthy controls for comparison to better understand the negative impact of symptomatic knee OA.

Diagnosing deviations and predicting faults is an important task, especially given recent advances related to Internet of Things. However, the majority of the efforts for diagnostics are still carried out by human experts in a time-consuming and expensive manner. One promising approach towards self-monitoring systems is based on the "wisdom of the crowd" idea, where malfunctioning equipments are detected by understanding the similarities and differences in the operation of several alike systems.

A fully autonomous fault detection, however, is not possible, since not all deviations or anomalies correspond to faulty behaviors; many can be explained by atypical usage or varying external conditions. In this work, we propose a method which gradually incorporates expert-provided feedback for more accurate self-monitoring. Our idea is to support model adaptation while allowing human feedback to persist over changes in data distribution, such as concept drift. © 2019 Association for Computing Machinery.

Unscheduled 30-day readmissions are a hallmark of Congestive Heart Failure (CHF) patients that pose significant health risks and escalate care cost. In order to reduce readmissions and curb the cost of care, it is important to initiate targeted intervention programs for patients at risk of readmission. This requires identifying high-risk patients at the time of discharge from hospital. Here, using real data from over 7,500 CHF patients hospitalized between 2012 and 2016 in Sweden, we built and tested a deep learning framework to predict 30-day unscheduled readmission. We present a cost-sensitive formulation of Long Short-Term Memory (LSTM) neural network using expert features and contextual embedding of clinical concepts. This study targets key elements of an Electronic Health Record (EHR) driven prediction model in a single framework: using both expert and machine derived features, incorporating sequential patterns and addressing the class imbalance problem. We show that the model with all key elements achieves a higher discrimination ability (AUC 0.77) compared to the rest. Additionally, we present a simple financial analysis to estimate annual savings if targeted interventions are offered to high risk patients. © 2019 The Authors