Showing papers in "BMC Medical Informatics and Decision Making in 2017"
TL;DR: Reducing within-patient repeats may be a promising target for reducing alert overrides and alert fatigue as clinicians became less likely to accept alerts as they received more of them, particularly more repeated alerts.
Abstract: Although alert fatigue is blamed for high override rates in contemporary clinical decision support systems, the concept of alert fatigue is poorly defined. We tested hypotheses arising from two possible alert fatigue mechanisms: (A) cognitive overload associated with amount of work, complexity of work, and effort distinguishing informative from uninformative alerts, and (B) desensitization from repeated exposure to the same alert over time. Retrospective cohort study using electronic health record data (both drug alerts and clinical practice reminders) from January 2010 through June 2013 from 112 ambulatory primary care clinicians. The cognitive overload hypotheses were that alert acceptance would be lower with higher workload (number of encounters, number of patients), higher work complexity (patient comorbidity, alerts per encounter), and more alerts low in informational value (repeated alerts for the same patient in the same year). The desensitization hypothesis was that, for newly deployed alerts, acceptance rates would decline after an initial peak. On average, one-quarter of drug alerts received by a primary care clinician, and one-third of clinical reminders, were repeats for the same patient within the same year. Alert acceptance was associated with work complexity and repeated alerts, but not with the amount of work. Likelihood of reminder acceptance dropped by 30% for each additional reminder received per encounter, and by 10% for each five percentage point increase in proportion of repeated reminders. The newly deployed reminders did not show a pattern of declining response rates over time, which would have been consistent with desensitization. Interestingly, nurse practitioners were 4 times as likely to accept drug alerts as physicians. Clinicians became less likely to accept alerts as they received more of them, particularly more repeated alerts. There was no evidence of an effect of workload per se, or of desensitization over time for a newly deployed alert. Reducing within-patient repeats may be a promising target for reducing alert overrides and alert fatigue.
319 citations
TL;DR: The general public was most interested in individualized and interactive health information by managing clinicians, they will highly benefit from using social media rather than Internet search tools, and the current state of health acquisition proves worrisome.
Abstract: On average, 570 million users, 93% in China’s first-tier cities, log on to WeChat every day. WeChat has become the most widely and frequently used social media in China, and has been profoundly integrated into the daily life of many Chinese people. A variety of health-related information may be found on WeChat. The objective of this study is to understand how the general public views the impact of the rapidly emerging social media on health information acquisition. A self-administered questionnaire was designed, distributed, collected, and analyzed utilizing the online survey tool Sojump. WeChat was adopted to randomly release the questionnaires using convenience sampling and collect the results after a certain amount of time. (1) A total of 1636 questionnaires (WeChat customers) were collected from 32 provinces. (2) The primary means by which respondents received health education was via the Internet (71.79%). Baidu and WeChat were the top 2 search tools utilized (90.71% and 28.30%, respectively). Only 12.41% of respondents were satisfied with their online health information search. (3) Almost all had seen (98.35%) or read (97.68%) health information; however, only 14.43% believed that WeChat health information could improve health. Nearly one-third frequently received and read health information through WeChat. WeChat was selected (63.26%) as the most expected means for obtaining health information. (4) The major concerns regarding health information through WeChat included the following: excessively homogeneous information, the lack of a guarantee of professionalism, and the presence of advertisements. (5) Finally, the general public was most interested in individualized and interactive health information by managing clinicians, they will highly benefit from using social media rather than Internet search tools. The current state of health acquisition proves worrisome. The public has a high chance to access health information via WeChat. The growing popularity of interactive social platforms (e.g. WeChat) presents a variety of challenges and opportunities with respect to public health acquisition.
204 citations
TL;DR: The size of literature in m-Health showed a noticeable increase in the past decade, and given the large volume of citations received in this field, it is expected that applications of m- health will be seen into various health aspects and health services.
Abstract: The advancement of mobile technology had positively influenced healthcare services An emerging subfield of mobile technology is mobile health (m-Health) in which mobile applications are used for health purposes The aim of this study was to analyze and assess literature published in the field of m-Health SciVerse Scopus was used to retrieve literature in m-Health The study period was set from 2006 to 2016 ArcGIS 101 was used to present geographical distribution of publications while VOSviewer was used for data visualization Growth of publications, citation analysis, and research productivity were presented using standard bibliometric indicators During the study period, a total of 5465 documents were published, giving an average of 4968 documents per year The h-index of retrieved documents was 81 Core keywords used in literature pertaining to m-Health included diabetes mellitus, adherence, and obesity among others Relative growth rate and doubling time of retrieved literature were stable from 2009 to 2015 indicating exponential growth of literature in this field A total of 4638 (849%) documents were multi-authored with a mean collaboration index of 41 authors per article The United States of America ranked first in productivity with 1926 (352%) published documents India ranked sixth with 183 (33%) documents while China ranked seventh with 155(28%) documents VA Medical Center was the most prolific organization/institution while Journal of Medical Internet Research was the preferred journal for publications in the field of m-Health Top cited articles in the field of m-Health included the use of mobile technology in improving adherence in HIV patients, weight loss, and improving glycemic control in diabetic patients The size of literature in m-Health showed a noticeable increase in the past decade Given the large volume of citations received in this field, it is expected that applications of m-Health will be seen into various health aspects and health services Research in m-Health needs to be encouraged, particularly in the fight against AIDS, poor medication adherence, glycemic control in Africa and other low income world regions where technology can improve health services and decrease disease burden
167 citations
TL;DR: This paper comprehensively investigates the performance of LSTM (long-short term memory), a representative variant of RNN, on clinical entity recognition and protected health information recognition, and shows that L STM outperforms traditional machine learning methods that suffer from fussy feature engineering.
Abstract: Entity recognition is one of the most primary steps for text analysis and has long attracted considerable attention from researchers. In the clinical domain, various types of entities, such as clinical entities and protected health information (PHI), widely exist in clinical texts. Recognizing these entities has become a hot topic in clinical natural language processing (NLP), and a large number of traditional machine learning methods, such as support vector machine and conditional random field, have been deployed to recognize entities from clinical texts in the past few years. In recent years, recurrent neural network (RNN), one of deep learning methods that has shown great potential on many problems including named entity recognition, also has been gradually used for entity recognition from clinical texts. In this paper, we comprehensively investigate the performance of LSTM (long-short term memory), a representative variant of RNN, on clinical entity recognition and protected health information recognition. The LSTM model consists of three layers: input layer – generates representation of each word of a sentence; LSTM layer – outputs another word representation sequence that captures the context information of each word in this sentence; Inference layer – makes tagging decisions according to the output of LSTM layer, that is, outputting a label sequence. Experiments conducted on corpora of the 2010, 2012 and 2014 i2b2 NLP challenges show that LSTM achieves highest micro-average F1-scores of 85.81% on the 2010 i2b2 medical concept extraction, 92.29% on the 2012 i2b2 clinical event detection, and 94.37% on the 2014 i2b2 de-identification, which is considerably competitive with other state-of-the-art systems. LSTM that requires no hand-crafted feature has great potential on entity recognition from clinical texts. It outperforms traditional machine learning methods that suffer from fussy feature engineering. A possible future direction is how to integrate knowledge bases widely existing in the clinical domain into LSTM, which is a case of our future work. Moreover, how to use LSTM to recognize entities in specific formats is also another possible future direction.
135 citations
TL;DR: This study shows that a supervised learning-based NLP approach is useful to develop medical subdomain classifiers, and the deep learning algorithm with distributed word representation yields better performance yet shallow learning algorithms with the word and concept representation achieves comparable performance with better clinical interpretability.
Abstract: The medical subdomain of a clinical note, such as cardiology or neurology, is useful content-derived metadata for developing machine learning downstream applications. To classify the medical subdomain of a note accurately, we have constructed a machine learning-based natural language processing (NLP) pipeline and developed medical subdomain classifiers based on the content of the note. We constructed the pipeline using the clinical NLP system, clinical Text Analysis and Knowledge Extraction System (cTAKES), the Unified Medical Language System (UMLS) Metathesaurus, Semantic Network, and learning algorithms to extract features from two datasets — clinical notes from Integrating Data for Analysis, Anonymization, and Sharing (iDASH) data repository (n = 431) and Massachusetts General Hospital (MGH) (n = 91,237), and built medical subdomain classifiers with different combinations of data representation methods and supervised learning algorithms. We evaluated the performance of classifiers and their portability across the two datasets. The convolutional recurrent neural network with neural word embeddings trained-medical subdomain classifier yielded the best performance measurement on iDASH and MGH datasets with area under receiver operating characteristic curve (AUC) of 0.975 and 0.991, and F1 scores of 0.845 and 0.870, respectively. Considering better clinical interpretability, linear support vector machine-trained medical subdomain classifier using hybrid bag-of-words and clinically relevant UMLS concepts as the feature representation, with term frequency-inverse document frequency (tf-idf)-weighting, outperformed other shallow learning classifiers on iDASH and MGH datasets with AUC of 0.957 and 0.964, and F1 scores of 0.932 and 0.934 respectively. We trained classifiers on one dataset, applied to the other dataset and yielded the threshold of F1 score of 0.7 in classifiers for half of the medical subdomains we studied. Our study shows that a supervised learning-based NLP approach is useful to develop medical subdomain classifiers. The deep learning algorithm with distributed word representation yields better performance yet shallow learning algorithms with the word and concept representation achieves comparable performance with better clinical interpretability. Portable classifiers may also be used across datasets from different institutions.
120 citations
TL;DR: Choice of a smoking cessation or alcohol reduction app may be influenced by its immediate look and feel, ‘social proof’ and titles that appear realistic, which may enhance motivation, autonomy, personal relevance and credibility.
Abstract: Public health organisations such as the National Health Service in the United Kingdom and the National Institutes of Health in the United States provide access to online libraries of publicly endorsed smartphone applications (apps); however, there is little evidence that users rely on this guidance. Rather, one of the most common methods of finding new apps is to search an online store. As hundreds of smoking cessation and alcohol-related apps are currently available on the market, smokers and drinkers must actively choose which app to download prior to engaging with it. The influences on this choice are yet to be identified. This study aimed to investigate 1) design features that shape users’ choice of smoking cessation or alcohol reduction apps, and 2) design features judged to be important for engagement. Adult smokers (n = 10) and drinkers (n = 10) interested in using an app to quit/cut down were asked to search an online store to identify and explore a smoking cessation or alcohol reduction app of their choice whilst thinking aloud. Semi-structured interview techniques were used to allow participants to elaborate on their statements. An interpretivist theoretical framework informed the analysis. Verbal reports were audio recorded, transcribed verbatim and analysed using inductive thematic analysis. Participants chose apps based on their immediate look and feel, quality as judged by others’ ratings and brand recognition (‘social proof’), and titles judged to be realistic and relevant. Monitoring and feedback, goal setting, rewards and prompts were identified as important for engagement, fostering motivation and autonomy. Tailoring of content, a non-judgmental communication style, privacy and accuracy were viewed as important for engagement, fostering a sense of personal relevance and trust. Sharing progress on social media and the use of craving management techniques in social settings were judged not to be engaging because of concerns about others’ negative reactions. Choice of a smoking cessation or alcohol reduction app may be influenced by its immediate look and feel, ‘social proof’ and titles that appear realistic. Design features that enhance motivation, autonomy, personal relevance and credibility may be important for engagement.
98 citations
TL;DR: These efforts on sentiment analysis for newly approved HPV vaccines provide an automatic and instant way to extract public opinion and understand the concerns on Twitter and can provide a feedback to public health professionals to monitor online public response.
Abstract: As one of the serious public health issues, vaccination refusal has been attracting more and more attention, especially for newly approved human papillomavirus (HPV) vaccines. Understanding public opinion towards HPV vaccines, especially concerns on social media, is of significant importance for HPV vaccination promotion. In this study, we leveraged a hierarchical machine learning based sentiment analysis system to extract public opinions towards HPV vaccines from Twitter. English tweets containing HPV vaccines-related keywords were collected from November 2, 2015 to March 28, 2016. Manual annotation was done to evaluate the performance of the system on the unannotated tweets corpus. Followed time series analysis was applied to this corpus to track the trends of machine-deduced sentiments and their associations with different days of the week. The evaluation of the unannotated tweets corpus showed that the micro-averaging F scores have reached 0.786. The learning system deduced the sentiment labels for 184,214 tweets in the collected unannotated tweets corpus. Time series analysis identified a coincidence between mainstream outcome and Twitter contents. A weak trend was found for “Negative” tweets that decreased firstly and began to increase later; an opposite trend was identified for “Positive” tweets. Tweets that contain the worries on efficacy for HPV vaccines showed a relative significant decreasing trend. Strong associations were found between some sentiments (“Positive”, “Negative”, “Negative-Safety” and “Negative-Others”) with different days of the week. Our efforts on sentiment analysis for newly approved HPV vaccines provide us an automatic and instant way to extract public opinion and understand the concerns on Twitter. Our approaches can provide a feedback to public health professionals to monitor online public response, examine the effectiveness of their HPV vaccination promotion strategies and adjust their promotion plans.
77 citations
TL;DR: The study shows the potential of machine learning methods for predicting all-cause mortality using cardiorespiratory fitness data and how the various techniques differ in terms of capabilities of predicting medical outcomes is compared.
Abstract: Prior studies have demonstrated that cardiorespiratory fitness (CRF) is a strong marker of cardiovascular health. Machine learning (ML) can enhance the prediction of outcomes through classification techniques that classify the data into predetermined categories. The aim of this study is to present an evaluation and comparison of how machine learning techniques can be applied on medical records of cardiorespiratory fitness and how the various techniques differ in terms of capabilities of predicting medical outcomes (e.g. mortality). We use data of 34,212 patients free of known coronary artery disease or heart failure who underwent clinician-referred exercise treadmill stress testing at Henry Ford Health Systems Between 1991 and 2009 and had a complete 10-year follow-up. Seven machine learning classification techniques were evaluated: Decision Tree (DT), Support Vector Machine (SVM), Artificial Neural Networks (ANN), Naive Bayesian Classifier (BC), Bayesian Network (BN), K-Nearest Neighbor (KNN) and Random Forest (RF). In order to handle the imbalanced dataset used, the Synthetic Minority Over-Sampling Technique (SMOTE) is used. Two set of experiments have been conducted with and without the SMOTE sampling technique. On average over different evaluation metrics, SVM Classifier has shown the lowest performance while other models like BN, BC and DT performed better. The RF classifier has shown the best performance (AUC = 0.97) among all models trained using the SMOTE sampling. The results show that various ML techniques can significantly vary in terms of its performance for the different evaluation metrics. It is also not necessarily that the more complex the ML model, the more prediction accuracy can be achieved. The prediction performance of all models trained with SMOTE is much better than the performance of models trained without SMOTE. The study shows the potential of machine learning methods for predicting all-cause mortality using cardiorespiratory fitness data.
66 citations
TL;DR: A patient-centred design approach was designed to support patients suffering from chronic obstructive pulmonary disease in self-managing their condition, resulting in high compliance with self-monitoring over a prolonged period of time.
Abstract: Recent telehealth studies have demonstrated minor impact on patients affected by long-term conditions. The use of technology does not guarantee the compliance required for sustained collection of high-quality symptom and physiological data. Remote monitoring alone is not sufficient for successful disease management. A patient-centred design approach is needed in order to allow the personalisation of interventions and encourage the completion of daily self-management tasks. A digital health system was designed to support patients suffering from chronic obstructive pulmonary disease in self-managing their condition. The system includes a mobile application running on a consumer tablet personal computer and a secure backend server accessible to the health professionals in charge of patient management. The patient daily routine included the completion of an adaptive, electronic symptom diary on the tablet, and the measurement of oxygen saturation via a wireless pulse oximeter. The design of the system was based on a patient-centred design approach, informed by patient workshops. One hundred and ten patients in the intervention arm of a randomised controlled trial were subsequently given the tablet computer and pulse oximeter for a 12-month period. Patients were encouraged, but not mandated, to use the digital health system daily. The average used was 6.0 times a week by all those who participated in the full trial. Three months after enrolment, patients were able to complete their symptom diary and oxygen saturation measurement in less than 1 m 40s (96% of symptom diaries). Custom algorithms, based on the self-monitoring data collected during the first 50 days of use, were developed to personalise alert thresholds. Strategies and tools aimed at refining a digital health intervention require iterative use to enable convergence on an optimal, usable design. ‘Continuous improvement’ allowed feedback from users to have an immediate impact on the design of the system (e.g., collection of quality data), resulting in high compliance with self-monitoring over a prolonged period of time (12-month). Health professionals were prompted by prioritisation algorithms to review patient data, which led to their regular use of the remote monitoring website throughout the trial. Trial registration: ISRCTN40367841
. Registered 17/10/2012.
64 citations
TL;DR: More than half of the health workers working at government health institutions of East Gojjam were poor health information users compared with the findings of others studies, suggesting a comprehensive training, supportive supervision, and regular feedback are highly recommended for improving routine health information utilization among health workers at governmenthealth facilities.
Abstract: Using reliable information from routine health information systems over time is an important aid to improving health outcomes, tackling disparities, enhancing efficiency, and encouraging innovation. In Ethiopia, routine health information utilization for enhancing performance is poor among health workers, especially at the peripheral levels of health facilities. Therefore, this study aimed to assess routine health information system utilization and associated factors among health workers at government health institutions in East Gojjam Zone, Northwest Ethiopia. An institution based cross-sectional study was conducted at government health institutions of East Gojjam Zone, Northwest Ethiopia from April to May, 2013. A total of 668 health workers were selected from government health institutions, using the cluster sampling technique. Data collected using a standard structured and self-administered questionnaire and an observational checklist were cleaned, coded, and entered into Epi-info version 3.5.3, and transferred into SPSS version 20 for further statistical analysis. Variables with a p-value of less than 0.05 at multiple logistic regression analysis were considered statistically significant factors for the utilization of routine health information systems. The study revealed that 45.8% of the health workers had a good level of routine health information utilization. HMIS training [AOR = 2.72, 95% CI: 1.60, 4.62], good data analysis skills [AOR = 6.40, 95%CI: 3.93, 10.37], supervision [AOR = 2.60, 95% CI: 1.42, 4.75], regular feedback [AOR = 2.20, 95% CI: 1.38, 3.51], and favorable attitude towards health information utilization [AOR = 2.85, 95% CI: 1.78, 4.54] were found significantly associated with a good level of routine health information utilization. More than half of the health workers working at government health institutions of East Gojjam were poor health information users compared with the findings of others studies. HMIS training, data analysis skills, supervision, regular feedback, and favorable attitude were factors related to routine health information system utilization. Therefore, a comprehensive training, supportive supervision, and regular feedback are highly recommended for improving routine health information utilization among health workers at government health facilities.
63 citations
TL;DR: The MOTECH platform’s technological limitations in ‘pushing’ out voice messages highlights the need for more timely use of data to mitigate delivery challenges and improve exposure to health information.
Abstract: Despite the growing use of technology in the health sector, little evidence is available on the technological performance of mobile health programs nor on the willingness of target users to utilize these technologies as intended (behavioral performance). In this case study of the Mobile Technology for Health (MOTECH) program in Ghana, we assess the platform’s effectiveness in delivering messages, along with user response across sites in five districts from 2011 to 2014. MOTECH is comprised of “Client Data Application" (CDA) which allows providers to digitize and track service delivery information for women and infants and “Mobile Midwife” (MM) which sends automated educational voice messages to the mobile phones of pregnant and postpartum women. Using a naturalist study design, we draw upon system generated data to evaluate message delivery, client engagement, and provider responsiveness to MOTECH over time and by level of facility. A total of 7,370 women were enrolled in MM during pregnancy and 14,867 women were enrolled postpa1rtum. While providers were able to register and upload patient-level health information using CDA, the majority of these uploads occurred in Community-based facilities versus Health Centers. For MM, 25% or less of expected messages were received by pregnant women, despite the majority (>77%) owning a private mobile phone. While over 80% of messages received by pregnant women were listened to, postpartum rates of listening declined over time. Only 25% of pregnant women received and listened to at least 1 first trimester message. By 6–12 months postpartum, less than 6% of enrolled women were exposed to at least one message. Caution should be exercised in assuming that digital health programs perform as intended. Evaluations should measure the technological, behavioral, health systems, and/or community factors which may lead to breaks in the impact pathway and influence findings on effectiveness. The MOTECH platform’s technological limitations in ‘pushing’ out voice messages highlights the need for more timely use of data to mitigate delivery challenges and improve exposure to health information. Alternative message delivery channels (USSD or SMS) could improve the platform’s ability to deliver messages but may not be appropriate for illiterate users. Not applicable.
TL;DR: Physicians believe in the usefulness of e-health, and professionals with previous experience with it are more open to its implementation and consider that the benefits of technology outweigh its possible difficulties and shortcomings.
Abstract: Technology has significantly changed the way health organizations operate. However, the role it plays in healthcare systems remains unclear. This aim of this study was to evaluate the opinion of physicians regarding e-health and determine what factors influence their opinion and describe the advantages, inconveniences and threats they may perceive by its use. A cross-sectional questionnaire-based study. A questionnaire which had been previously designed and validated by the authors was used to interview physicians from the Barcelona Medical Association. 930 physicians were contacted by phone to participate in the study. Seven hundred sixty physicians responded to the questionnaire (response rate: 82%). The usefulness of telemedicine scored 7.4 (SD 1.8) on a scale from 1–10 (from the lowest to the highest) and the importance of the Internet in the workplace was 8.2 points (SD 1.8). Therapeutic compliance (7.0 -SD 1.8-) and patient health (7.0 -SD 1.7-) showed the best scores, and there were differences between professionals who had and had not previously participated in a telemedicine project (p < 0.05). The multivariate regression model explained the 41% of the variance for 7 factors: participation in telemedicine project (p < 0.001), quality of clinical practice (p < 0.001), patient health (p < 0.001), professional workload (p = 0.005), ease-of-use of electronic device (p = 0.007), presence of incentives for telemedicine (p = 0.011) and patient preference for in-person visits (p = 0.05). Physicians believe in the usefulness of e-health. Professionals with previous experience with it are more open to its implementation and consider that the benefits of technology outweigh its possible difficulties and shortcomings. Physicians demanded projects with appropriate funding and technology, as well as specific training to improve their technological abilities. The relationship of users with technology differs according to their personal or professional life. Although a 2.0 philosophy has been incorporated into many aspects of our lives, healthcare systems still have a long way to go in order to adapt to this new understanding of the relationship between patients and their health.
TL;DR: Examining word2vec’s ability in deriving semantic relatedness and similarity between biomedical terms from large publication data finds that increasing the size of datasets can result in the identification of more relations of biomedical terms even though it does not guarantee better precision.
Abstract: Understanding semantic relatedness and similarity between biomedical terms has a great impact on a variety of applications such as biomedical information retrieval, information extraction, and recommender systems. The objective of this study is to examine word2vec’s ability in deriving semantic relatedness and similarity between biomedical terms from large publication data. Specifically, we focus on the effects of recency, size, and section of biomedical publication data on the performance of word2vec. We download abstracts of 18,777,129 articles from PubMed and 766,326 full-text articles from PubMed Central (PMC). The datasets are preprocessed and grouped into subsets by recency, size, and section. Word2vec models are trained on these subtests. Cosine similarities between biomedical terms obtained from the word2vec models are compared against reference standards. Performance of models trained on different subsets are compared to examine recency, size, and section effects. Models trained on recent datasets did not boost the performance. Models trained on larger datasets identified more pairs of biomedical terms than models trained on smaller datasets in relatedness task (from 368 at the 10% level to 494 at the 100% level) and similarity task (from 374 at the 10% level to 491 at the 100% level). The model trained on abstracts produced results that have higher correlations with the reference standards than the one trained on article bodies (i.e., 0.65 vs. 0.62 in the similarity task and 0.66 vs. 0.59 in the relatedness task). However, the latter identified more pairs of biomedical terms than the former (i.e., 344 vs. 498 in the similarity task and 339 vs. 503 in the relatedness task). Increasing the size of dataset does not always enhance the performance. Increasing the size of datasets can result in the identification of more relations of biomedical terms even though it does not guarantee better precision. As summaries of research articles, compared with article bodies, abstracts excel in accuracy but lose in coverage of identifiable relations.
TL;DR: The proposed Bayesian network based model not only identifies known and suspected high PU risk factors, but also substantially increases sensitivity of the prediction - nearly three times higher comparing to logistical regression models - without sacrificing the overall accuracy.
Abstract: We develop predictive models enabling clinicians to better understand and explore patient clinical data along with risk factors for pressure ulcers in intensive care unit patients from electronic health record data. Identifying accurate risk factors of pressure ulcers is essential to determining appropriate prevention strategies; in this work we examine medication, diagnosis, and traditional Braden pressure ulcer assessment scale measurements as patient features. In order to predict pressure ulcer incidence and better understand the structure of related risk factors, we construct Bayesian networks from patient features. Bayesian network nodes (features) and edges (conditional dependencies) are simplified with statistical network techniques. Upon reviewing a network visualization of our model, our clinician collaborators were able to identify strong relationships between risk factors widely recognized as associated with pressure ulcers. We present a three-stage framework for predictive analysis of patient clinical data: 1) Developing electronic health record feature extraction functions with assistance of clinicians, 2) simplifying features, and 3) building Bayesian network predictive models. We evaluate all combinations of Bayesian network models from different search algorithms, scoring functions, prior structure initializations, and sets of features. From the EHRs of 7,717 ICU patients, we construct Bayesian network predictive models from 86 medication, diagnosis, and Braden scale features. Our model not only identifies known and suspected high PU risk factors, but also substantially increases sensitivity of the prediction - nearly three times higher comparing to logistical regression models - without sacrificing the overall accuracy. We visualize a representative model with which our clinician collaborators identify strong relationships between risk factors widely recognized as associated with pressure ulcers. Given the strong adverse effect of pressure ulcers on patients and the high cost for treating pressure ulcers, our Bayesian network based model provides a novel framework for significantly improving the sensitivity of the prediction model. Thus, when the model is deployed in a clinical setting, the caregivers can suitably respond to conditions likely associated with pressure ulcer incidence.
TL;DR: An interactive tool aimed at app developers, summarising key features of the policy environment and highlighting legislative, industry and professional standards around seven relevant domains is developed, anticipating that mental health apps developed in accordance with this tool will be more likely to conform to regulatory requirements, protect consumer privacy, protect consumers finances, and deliver health benefit.
Abstract: Apps targeted at health and wellbeing sit in a rapidly growing industry associated with widespread optimism about their potential to deliver accessible and cost-effective healthcare. App developers might not be aware of all the regulatory requirements and best practice principles are emergent. Health apps are regulated in order to minimise their potential for harm due to, for example, loss of personal health privacy, financial costs, and health harms from delayed or unnecessary diagnosis, monitoring and treatment. We aimed to produce a comprehensive guide to assist app developers in producing health apps that are legally compliant and in keeping with high professional standards of user protection. We conducted a case study analysis of the Australian and related international policy environment for mental health apps to identify relevant sectors, policy actors, and policy solutions. We identified 29 policies produced by governments and non-government organisations that provide oversight of health apps. In consultation with stakeholders, we developed an interactive tool targeted at app developers, summarising key features of the policy environment and highlighting legislative, industry and professional standards around seven relevant domains: privacy, security, content, promotion and advertising, consumer finances, medical device efficacy and safety, and professional ethics. We annotated this developer guidance tool with information about: the relevance of each domain; existing legislative and non-legislative guidance; critiques of existing policy; recommendations for developers; and suggestions for other key stakeholders. We anticipate that mental health apps developed in accordance with this tool will be more likely to conform to regulatory requirements, protect consumer privacy, protect consumer finances, and deliver health benefit; and less likely to attract regulatory penalties, offend consumers and communities, mislead consumers, or deliver health harms. We encourage government, industry and consumer organisations to use and publicise the tool.
TL;DR: Improved mortality prediction at hospital discharge after first MI is important for identifying high-risk individuals eligible for intensified treatment and care and because of the superior national coverage, the best model can potentially be used to better differentiate new patients, allowing for improved targeting of limited resources.
Abstract: Machine learning algorithms hold potential for improved prediction of all-cause mortality in cardiovascular patients, yet have not previously been developed with high-quality population data. This study compared four popular machine learning algorithms trained on unselected, nation-wide population data from Sweden to solve the binary classification problem of predicting survival versus non-survival 2 years after first myocardial infarction (MI). This prospective national registry study for prognostic accuracy validation of predictive models used data from 51,943 complete first MI cases as registered during 6 years (2006–2011) in the national quality register SWEDEHEART/RIKS-HIA (90% coverage of all MIs in Sweden) with follow-up in the Cause of Death register (> 99% coverage). Primary outcome was AUROC (C-statistic) performance of each model on the untouched test set (40% of cases) after model development on the training set (60% of cases) with the full (39) predictor set. Model AUROCs were bootstrapped and compared, correcting the P-values for multiple comparisons with the Bonferroni method. Secondary outcomes were derived when varying sample size (1–100% of total) and predictor sets (39, 10, and 5) for each model. Analyses were repeated on 79,869 completed cases after multivariable imputation of predictors. A Support Vector Machine with a radial basis kernel developed on 39 predictors had the highest complete cases performance on the test set (AUROC = 0.845, PPV = 0.280, NPV = 0.966) outperforming Boosted C5.0 (0.845 vs. 0.841, P = 0.028) but not significantly higher than Logistic Regression or Random Forest. Models converged to the point of algorithm indifference with increased sample size and predictors. Using the top five predictors also produced good classifiers. Imputed analyses had slightly higher performance. Improved mortality prediction at hospital discharge after first MI is important for identifying high-risk individuals eligible for intensified treatment and care. All models performed accurately and similarly and because of the superior national coverage, the best model can potentially be used to better differentiate new patients, allowing for improved targeting of limited resources. Future research should focus on further model development and investigate possibilities for implementation.
TL;DR: This study demonstrates the feasibility of using a semantic content-based recommender system to enrich YouTube health videos by finding the majority of websites recommended by this system for health videos were relevant, based on ratings by health professionals.
Abstract: The Internet, and its popularity, continues to grow at an unprecedented pace. Watching videos online is very popular; it is estimated that 500 h of video are uploaded onto YouTube, a video-sharing service, every minute and that, by 2019, video formats will comprise more than 80% of Internet traffic. Health-related videos are very popular on YouTube, but their quality is always a matter of concern. One approach to enhancing the quality of online videos is to provide additional educational health content, such as websites, to support health consumers. This study investigates the feasibility of building a content-based recommender system that links health consumers to reputable health educational websites from MedlinePlus for a given health video from YouTube. The dataset for this study includes a collection of health-related videos and their available metadata. Semantic technologies (such as SNOMED-CT and Bio-ontology) were used to recommend health websites from MedlinePlus. A total of 26 healths professionals participated in evaluating 253 recommended links for a total of 53 videos about general health, hypertension, or diabetes. The relevance of the recommended health websites from MedlinePlus to the videos was measured using information retrieval metrics such as the normalized discounted cumulative gain and precision at K. The majority of websites recommended by our system for health videos were relevant, based on ratings by health professionals. The normalized discounted cumulative gain was between 46% and 90% for the different topics. Our study demonstrates the feasibility of using a semantic content-based recommender system to enrich YouTube health videos. Evaluation with end-users, in addition to healthcare professionals, will be required to identify the acceptance of these recommendations in a nonsimulated information-seeking context.
TL;DR: The main objective of this study was to explore and identify the multi-level (micro, meso and macro) factors affecting telemedicine utilization in Norway and provide a knowledge base useful to other countries which intend to implement teleMedicine or other digital health services into their healthcare systems.
Abstract: Norway has a long history of using telemedicine, especially for geographical reasons. Despite the availability of promising telemedicine applications and the implementation of national initiatives and policies, the sustainability and scaling-up of telemedicine in the health system is still far from accomplished. The main objective of this study was to explore and identify the multi-level (micro, meso and macro) factors affecting telemedicine utilization in Norway. We used a mixed methods approach. Data from a national registry were collected to analyze the use of outpatient visits and telemedicine contacts in Norway from 2009 to 2015. Interviews with key stakeholders at national, regional and local level helped complete and contextualize the data analysis and explore the main issues affecting the use of telemedicine by health authorities and hospitals. Relevant national documents were also used to support, contradict, contextualize or clarify information and data. Telemedicine use in Norway from 2009 to 2015 remained very low, not exceeding 0.5% of total outpatient activity at regional level and 0.1% at national level. All four regions used telemedicine. Of the 29 hospitals, 24 used it at least once over the 7-year period. Telemedicine was not used regularly everywhere, with some hospitals using it sporadically. Telemedicine was mostly used in selected specialties, including rehabilitation, neurosurgery, skin and venereal diseases. Three major themes affecting implementation and utilization of telemedicine in Norway emerged: (i) governance and strategy; (ii) organizational and professional dimensions; (iii) economic and financial dimensions. For each theme, a number of factors and challenges faced at different health care levels were identified. This study allowed shedding light on multi-level and interdependent factors affecting utilization of telemedicine in Norway. The identification of the main implementation and utilization challenges might support decision makers and practitioners in the successful scaling-up of telemedicine. This work provides a knowledge base useful to other countries which intend to implement telemedicine or other digital health services into their healthcare systems.
TL;DR: This study found evidence of AB omission and commission errors in e-prescribing and the impact of task complexity and interruptions on AB, and found verification of CDS alerts is key to avoiding AB errors.
Abstract: Clinical decision support (CDS) in e-prescribing can improve safety by alerting potential errors, but introduces new sources of risk. Automation bias (AB) occurs when users over-rely on CDS, reducing vigilance in information seeking and processing. Evidence of AB has been found in other clinical tasks, but has not yet been tested with e-prescribing. This study tests for the presence of AB in e-prescribing and the impact of task complexity and interruptions on AB. One hundred and twenty students in the final two years of a medical degree prescribed medicines for nine clinical scenarios using a simulated e-prescribing system. Quality of CDS (correct, incorrect and no CDS) and task complexity (low, low + interruption and high) were varied between conditions. Omission errors (failure to detect prescribing errors) and commission errors (acceptance of false positive alerts) were measured. Compared to scenarios with no CDS, correct CDS reduced omission errors by 38.3% (p < .0001, n = 120), 46.6% (p < .0001, n = 70), and 39.2% (p < .0001, n = 120) for low, low + interrupt and high complexity scenarios respectively. Incorrect CDS increased omission errors by 33.3% (p < .0001, n = 120), 24.5% (p < .009, n = 82), and 26.7% (p < .0001, n = 120). Participants made commission errors, 65.8% (p < .0001, n = 120), 53.5% (p < .0001, n = 82), and 51.7% (p < .0001, n = 120). Task complexity and interruptions had no impact on AB. This study found evidence of AB omission and commission errors in e-prescribing. Verification of CDS alerts is key to avoiding AB errors. However, interventions focused on this have had limited success to date. Clinicians should remain vigilant to the risks of CDS failures and verify CDS.
TL;DR: In this article, the authors analyzed over 45,000 Facebook posts from 72 Facebook accounts belonging to 24 U.S. federal health agencies and found that agencies and accounts vary widely in their usage of social media and activity they generate.
Abstract: It is becoming increasingly common for individuals and organizations to use social media platforms such as Facebook. These are being used for a wide variety of purposes including disseminating, discussing and seeking health related information. U.S. Federal health agencies are leveraging these platforms to ‘engage’ social media users to read, spread, promote and encourage health related discussions. However, different agencies and their communications get varying levels of engagement. In this study we use statistical models to identify factors that associate with engagement. We analyze over 45,000 Facebook posts from 72 Facebook accounts belonging to 24 health agencies. Account usage, user activity, sentiment and content of these posts are studied. We use the hurdle regression model to identify factors associated with the level of engagement and Cox proportional hazards model to identify factors associated with duration of engagement. In our analysis we find that agencies and accounts vary widely in their usage of social media and activity they generate. Statistical analysis shows, for instance, that Facebook posts with more visual cues such as photos or videos or those which express positive sentiment generate more engagement. We further find that posts on certain topics such as occupation or organizations negatively affect the duration of engagement. We present the first comprehensive analyses of engagement with U.S. Federal health agencies on Facebook. In addition, we briefly compare and contrast findings from this study to our earlier study with similar focus but on Twitter to show the robustness of our methods.
TL;DR: Most physical and psychiatric disorders were more common in the BPD patients than in the control patients, and clinicians caring for people with BPD should be aware of possible comorbidities.
Abstract: Borderline personality disorder (BPD) is a complex clinical state with highly polymorphic symptoms and signs. Studies have demonstrated that people with a BPD diagnosis are likely to have numerous co-occurring psychiatric disorders and physical comorbidities. The aim of our study was to obtain further insight about the associations among comorbidities of BPD and to demonstrate the practicality of using association rule mining (ARM) technique in clinical databases. A retrospective case–control study was conducted on information of 1460 patients (292 BPD patients and 1168 control patients) selected from the Taiwan National Health Insurance Research Database. Information on physical and psychiatric comorbidities, which were diagnosed within 3 years before and after enrollment, was collected. A logistic regression model was used to calculate the odds ratios of comorbidities between patients with and without BPD. ARM technique was used to study the associations of BPD and two or more psychiatric comorbidities. We classified physical comorbidities into 13 categories according to the International Classification of Diseases, Ninth Revision, Clinical Modification system, and the results indicated that the 12 categories were more common in the BPD patients than in the control patients (except congenital anomalies). However, psychiatric comorbidities, including depressive disorder, bipolar disorder, anxiety disorder, sleep disorder, substance use disorder, and mental retardation were more common in the BPD patients than in the control patients. Furthermore, the associations of BPD and two or more comorbidities were evaluated. Most physical and psychiatric disorders were more common in the BPD patients than in the control patients. Because the failure to remit from BPD is associated with suffering from chronic physical conditions and because psychiatric comorbidities may lead to delays in diagnosis of BPD, clinicians caring for people with BPD should be aware of possible comorbidities.
TL;DR: A new utility-preserving anonymization method for privacy preserving data publishing (PPDP) and an anonymization algorithm using the proposed method are proposed that show the lower information loss than the existing method.
Abstract: Publishing raw electronic health records (EHRs) may be considered as a breach of the privacy of individuals because they usually contain sensitive information. A common practice for the privacy-preserving data publishing is to anonymize the data before publishing, and thus satisfy privacy models such as k-anonymity. Among various anonymization techniques, generalization is the most commonly used in medical/health data processing. Generalization inevitably causes information loss, and thus, various methods have been proposed to reduce information loss. However, existing generalization-based data anonymization methods cannot avoid excessive information loss and preserve data utility. We propose a utility-preserving anonymization for privacy preserving data publishing (PPDP). To preserve data utility, the proposed method comprises three parts: (1) utility-preserving model, (2) counterfeit record insertion, (3) catalog of the counterfeit records. We also propose an anonymization algorithm using the proposed method. Our anonymization algorithm applies full-domain generalization algorithm. We evaluate our method in comparison with existence method on two aspects, information loss measured through various quality metrics and error rate of analysis result. With all different types of quality metrics, our proposed method show the lower information loss than the existing method. In the real-world EHRs analysis, analysis results show small portion of error between the anonymized data through the proposed method and original data. We propose a new utility-preserving anonymization method and an anonymization algorithm using the proposed method. Through experiments on various datasets, we show that the utility of EHRs anonymized by the proposed method is significantly better than those anonymized by previous approaches.
TL;DR: This approach identified patients with MS early in the course of their disease which could potentially shorten the time to diagnosis, and could also be applied to other diseases often missed by primary care providers such as cancer.
Abstract: Diagnostic accuracy might be improved by algorithms that searched patients’ clinical notes in the electronic health record (EHR) for signs and symptoms of diseases such as multiple sclerosis (MS). The focus this study was to determine if patients with MS could be identified from their clinical notes prior to the initial recognition by their healthcare providers. An MS-enriched cohort of patients with well-established MS (n = 165) and controls (n = 545), was generated from the adult outpatient clinic. A random sample cohort was generated from randomly selected patients (n = 2289) from the same adult outpatient clinic, some of whom had MS (n = 16). Patients’ notes were extracted from the data warehouse and signs and symptoms mapped to UMLS terms using MedLEE. Approximately 1000 MS-related terms occurred significantly more frequently in MS patients’ notes than controls’. Synonymous terms were manually clustered into 50 buckets and used as classification features. Patients were classified as MS or not using Naive Bayes classification. Classification of patients known to have MS using notes of the MS-enriched cohort entered after the initial ICD9[MS] code yielded an ROC AUC, sensitivity, and specificity of 0.90 [0.87-0.93], 0.75[0.66-0.82], and 0.91 [0.87-0.93], respectively. Similar classification accuracy was achieved using the notes from the random sample cohort. Classification of patients not yet known to have MS using notes of the MS-enriched cohort entered before the initial ICD9[MS] documentation identified 40% [23–59%] as having MS. Manual review of the EHR of 45 patients of the random sample cohort classified as having MS but lacking an ICD9[MS] code identified four who might have unrecognized MS. Diagnostic accuracy might be improved by mining patients’ clinical notes for signs and symptoms of specific diseases using NLP. Using this approach, we identified patients with MS early in the course of their disease which could potentially shorten the time to diagnosis. This approach could also be applied to other diseases often missed by primary care providers such as cancer. Whether implementing computerized diagnostic support ultimately shortens the time from earliest symptoms to formal recognition of the disease remains to be seen.
TL;DR: A new adoption model using as a starting point the extended Unified Theory of Acceptance and Use of Technology (UTAUT2) by incorporating the Concern for Information Privacy (CFIP) framework was applied and identified critical factors for the adoption of EHR portals and significant differences between the countries.
Abstract: This study’s goal is to understand the factors that drive individuals to adopt Electronic Health Record (EHR) portals and to estimate if there are differences between countries with different healthcare models. We applied a new adoption model using as a starting point the extended Unified Theory of Acceptance and Use of Technology (UTAUT2) by incorporating the Concern for Information Privacy (CFIP) framework. To evaluate the research model we used the partial least squares (PLS) – structural equation modelling (SEM) approach. An online questionnaire was administrated in the United States (US) and Europe (Portugal). We collected 597 valid responses. The statistically significant factors of behavioural intention are performance expectancy (
$$ \widehat{\beta} $$
total = 0.285; P < 0.01), effort expectancy (
$$ \widehat{\beta} $$
total = 0.160; P < 0.01), social influence (
$$ \widehat{\beta} $$
total = 0.198; P < 0.01), hedonic motivation (
$$ \widehat{\beta} $$
total = −0.141; P < 0.01), price value (
$$ \widehat{\beta} $$
total = 0.152; P < 0.01), and habit (
$$ \widehat{\beta} $$
total = 0.255; P < 0.01). The predictors of use behaviour are habit (
$$ \widehat{\beta} $$
total = 0.145; P < 0.01), and behavioural intention (
$$ \widehat{\beta} $$
total = 0.480; P < 0.01). Social influence, hedonic motivation, and price value are only predictors in the US group. The model explained 53% of the variance in behavioural intention and 36% of the variance in use behaviour. Our study identified critical factors for the adoption of EHR portals and significant differences between the countries. Confidentiality issues do not seem to influence acceptance. The EHR portals usage patterns are significantly higher in US compared to Portugal.
TL;DR: Despite an increasingly complex client population in the home and community care sectors, the results from this work indicate that data collected using the RAI-HC and the CHA are of an overall quality that may be trusted when used to inform decision-making at the organizational- or policy-level.
Abstract: The aim of this project is to describe the quality of assessment data regularly collected in home and community, with techniques adapted from an evaluation of the quality of long-term care data in Canada. Data collected using the Resident Assessment Instrument – Home Care (RAI-HC) in Ontario and British Columbia (BC) as well as the interRAI Community Health Assessment (CHA) in Ontario were analyzed using descriptive statistics, Pearson’s r correlation, and Cronbach’s alpha in order to assess trends in population characteristics, convergent validity, and scale reliability. Results indicate that RAI-HC data from Ontario and BC behave in a consistent manner, with stable trends in internal consistency providing evidence of good reliability (alpha values range from 0.72-0.94, depending on the scale and province). The associations between various scales, such as those reflecting functional status and cognition, were found to be as expected and stable over time within each setting (r values range from 0.42-0.45 in Ontario and 0.41-0.43 in BC). These trends in convergent validity demonstrate that constructs in the data behave as they should, providing evidence of good data quality. In most cases, CHA data quality matches that of RAI-HC data quality and shows evidence of good validity and reliability. The findings are comparable to the findings observed in the evaluation of data from the long-term care sector. Despite an increasingly complex client population in the home and community care sectors, the results from this work indicate that data collected using the RAI-HC and the CHA are of an overall quality that may be trusted when used to inform decision-making at the organizational- or policy-level. High quality data and information are vital when used to inform steps taken to improve quality of care and enhance quality of life. This work also provides evidence that a method used to evaluate the quality of data obtained in the long-term care setting may be used to evaluate the quality of data obtained through community-based measures.
TL;DR: A secure protocol for the deduplication of horizontally partitioned datasets with deterministic record linkage algorithms is designed and implemented and demonstrated that it protects the privacy of individuals and data custodians under a semi-honest adversarial model.
Abstract: Techniques have been developed to compute statistics on distributed datasets without revealing private information except the statistical results. However, duplicate records in a distributed dataset may lead to incorrect statistical results. Therefore, to increase the accuracy of the statistical analysis of a distributed dataset, secure deduplication is an important preprocessing step. We designed a secure protocol for the deduplication of horizontally partitioned datasets with deterministic record linkage algorithms. We provided a formal security analysis of the protocol in the presence of semi-honest adversaries. The protocol was implemented and deployed across three microbiology laboratories located in Norway, and we ran experiments on the datasets in which the number of records for each laboratory varied. Experiments were also performed on simulated microbiology datasets and data custodians connected through a local area network. The security analysis demonstrated that the protocol protects the privacy of individuals and data custodians under a semi-honest adversarial model. More precisely, the protocol remains secure with the collusion of up to N − 2 corrupt data custodians. The total runtime for the protocol scales linearly with the addition of data custodians and records. One million simulated records distributed across 20 data custodians were deduplicated within 45 s. The experimental results showed that the protocol is more efficient and scalable than previous protocols for the same problem. The proposed deduplication protocol is efficient and scalable for practical uses while protecting the privacy of patients and data custodians.
TL;DR: Users found the app to be a valuable tool for support, particularly for raising their awareness about their psychological health and for informing and guiding them through the healthcare system after diagnosis.
Abstract: Numerous mobile applications have been developed to support diabetes-self-management. However, the majority of these applications lack a theoretical foundation and the involvement of people with diabetes during development. The aim of this study was to develop and test a mobile application (app) supporting diabetes self-management among people with newly diagnosed type 2 diabetes using design thinking. The app was developed and tested in 2015 using a design-based research approach involving target users (individuals newly diagnosed with type 2 diabetes), research scientists, healthcare professionals, designers, and app developers. The research approach comprised three major phases: inspiration, ideation, and implementation. The first phase included observations of diabetes education and 12 in-depth interviews with users regarding challenges and needs related to living with diabetes. The ideation phrase consisted of four interactive workshops with users focusing on app needs, in which ideas were developed and prioritized. Finally, 14 users tested the app over 4 weeks; they were interviewed about usability and perceptions about the app as a support tool. A multifunctional app was useful for people with newly diagnosed type 2 diabetes. The final app comprised five major functions: overview of diabetes activities after diagnosis, recording of health data, reflection games and goal setting, knowledge games and recording of psychological data such as sleep, fatigue, and well-being. Users found the app to be a valuable tool for support, particularly for raising their awareness about their psychological health and for informing and guiding them through the healthcare system after diagnosis. The design thinking processes used in the development and implementation of the mobile health app were crucial to creating value for users. More attention should be paid to the training of professionals who introduce health apps. Trial registration: Danish Data Protection Agency: 2012-58-0004. Registered 6 February 2016.
TL;DR: Evaluating the performance of traditional classifiers for identifying patients with Systemic Lupus Erythematosus (SLE) in comparison with a newer Bayesian word vector method suggests that a shallow neural network with CUIs and random forests with both CUI and BOWs are the best classifier for this lupus phenotyping task.
Abstract: Identifying patients with certain clinical criteria based on manual chart review of doctors’ notes is a daunting task given the massive amounts of text notes in the electronic health records (EHR). This task can be automated using text classifiers based on Natural Language Processing (NLP) techniques along with pattern recognition machine learning (ML) algorithms. The aim of this research is to evaluate the performance of traditional classifiers for identifying patients with Systemic Lupus Erythematosus (SLE) in comparison with a newer Bayesian word vector method. We obtained clinical notes for patients with SLE diagnosis along with controls from the Rheumatology Clinic (662 total patients). Sparse bag-of-words (BOWs) and Unified Medical Language System (UMLS) Concept Unique Identifiers (CUIs) matrices were produced using NLP pipelines. These matrices were subjected to several different NLP classifiers: neural networks, random forests, naive Bayes, support vector machines, and Word2Vec inversion, a Bayesian inversion method. Performance was measured by calculating accuracy and area under the Receiver Operating Characteristic (ROC) curve (AUC) of a cross-validated (CV) set and a separate testing set. We calculated the accuracy of the ICD-9 billing codes as a baseline to be 90.00% with an AUC of 0.900, the shallow neural network with CUIs to be 92.10% with an AUC of 0.970, the random forest with BOWs to be 95.25% with an AUC of 0.994, the random forest with CUIs to be 95.00% with an AUC of 0.979, and the Word2Vec inversion to be 90.03% with an AUC of 0.905. Our results suggest that a shallow neural network with CUIs and random forests with both CUIs and BOWs are the best classifiers for this lupus phenotyping task. The Word2Vec inversion method failed to significantly beat the ICD-9 code classification, but yielded promising results. This method does not require explicit features and is more adaptable to non-binary classification tasks. The Word2Vec inversion is hypothesized to become more powerful with access to more data. Therefore, currently, the shallow neural networks and random forests are the desirable classifiers.
TL;DR: HITAM helped understand the likelihood that older people with LTCs would use HIT, but did not explain how this might result in improved self-management, and equipment design and organisational factors need to be considered in order to increase HIT acceptance among older people.
Abstract: Health information technology (HIT) may be used to improve care for increasing numbers of older people with long term conditions (LTCs) who make high demands on health and social care services. Despite its potential benefits for reducing disease exacerbations and hospitalisations, HIT home monitoring is not always accepted by patients. Using the Health Information Technology Acceptance Model (HITAM) this qualitative study examined the usefulness of the model for understanding acceptance of HIT in older people (≥60 years) participating in a RCT for older people with Chronic Obstructive Pulmonary Disease (COPD) and associated heart diseases (CHROMED). An instrumental, collective case study design was used with qualitative interviews of patients in the intervention arm of CHROMED. These were conducted at two time points, one shortly after installation of equipment and again at the end of (or withdrawal from) the study. We used Framework Analysis to examine how well the HITAM accounted for the data. Participants included 21 patients aged between 60–99 years and their partners or relatives where applicable. Additional concepts for the HITAM for older people included: concerns regarding health professional access and attachment; heightened illness anxiety and desire to avoid continuation of the ‘sick-role’. In the technology zone, HIT self-efficacy was associated with good organisational processes and informal support; while ease of use was connected to equipment design being suitable for older people. HIT perceived usefulness was related to establishing trends in health status, detecting early signs of infection and potential to self-manage. Due to limited feedback to users opportunities to self-manage were reduced. HITAM helped understand the likelihood that older people with LTCs would use HIT, but did not explain how this might result in improved self-management. In order to increase HIT acceptance among older people, equipment design and organisational factors need to be considered. ClinicalTrials.gov Identifier: NCT01960907
October 9 2013 (retrospectively registered) Clinical tRials fOr elderly patients with MultiplE Disease (CHROMED). Start date October 2012, end date March 2016. Date of enrolment of the first participant was February 2013.
TL;DR: Exposure to ICT and eHealth skills training are needed for adults with schizophrenia spectrum disorders from two distant European regions, where seHealth literacy is either moderate (Finnish group) or low (Greek group).
Abstract: Individuals with schizophrenia spectrum disorders use the Internet for general and health-related purposes. Their ability to find, understand, and apply the health information they acquire online in order to make appropriate health decisions – known as eHealth literacy – has never been investigated. The European agenda strives to limit health inequalities and enhance mental health literacy. Nevertheless, each European member state varies in levels of Internet use and online health information-seeking. This study aimed to examine computer/Internet use for general and health-related purposes, eHealth literacy, and attitudes toward computer/Internet among adults with schizophrenia spectrum disorders from two distant European regions. Data were collected from mental health services of psychiatric clinics in Finland (FI) and Greece (GR). A total of 229 patients (FI = 128, GR = 101) participated in the questionnaire survey. The data analysis included evaluation of frequencies and group comparisons with multiple linear and logistic regression models. The majority of Finnish participants were current Internet users (FI = 111, 87%, vs. GR = 33, 33%, P < .0001), while the majority of Greek participants had never used computers/Internet, mostly due to their perception that they do not need it. In both countries, more than half of Internet users used the Internet for health-related purposes (FI = 61, 55%, vs. GR = 20, 61%). The eHealth literacy of Internet users (previous and current Internet users) was found significantly higher in the Finnish group (FI: Mean = 27.05, SD 5.36; GR: Mean = 23.15, SD = 7.23, P < .0001) upon comparison with their Greek counterparts. For current Internet users, Internet use patterns were significantly different between country groups. When adjusting for gender, age, education and disease duration, country was a significant predictor of frequency of Internet use, eHealth literacy and Interest. The Finnish group of Internet users scored higher in eHealth literacy, while the Greek group of never Internet users had a higher Interest in computer/Internet. eHealth literacy is either moderate (Finnish group) or low (Greek group). Thus, exposure to ICT and eHealth skills training are needed for this population. Recommendations to improve the eHealth literacy and access to health information among these individuals are provided.