A multi-stage transfer learning strategy for diagnosing a class of rare laryngeal movement disorders.

BACKGROUND: It remains hard to directly apply deep learning-based methods to assist diagnosing essential tremor of voice (ETV) and abductor and adductor spasmodic dysphonia (ABSD and ADSD). One of the main challenges is that, as a class of rare laryngeal movement disorders (LMDs), there are limited available databases to be investigated. Another worthy explored research question is which above sub-disorder benefits most from diagnosis based on sustained phonations. The question is from the fact that sustained phonations can help detect pathological voice from healthy voice. METHOD: A transfer learning strategy is developed for LMD diagnosis with limited data, which consists of three fundamental parts. (1) An extra vocally healthy database from the International Dialects of English Archive (IDEA) is employed to pre-train a convolutional autoencoder. (2) The transferred proportion of the pre-trained encoder is explored. And its impact on LMD diagnosis is also evaluated, yielding a two-stage transfer model. (3) A third stage is designed following the initial two stages to embed information of pathological sustained phonation into the model. This stage verifies the different effects of applying sustained phonation on diagnosing the three sub-disorders, and helps boost the final diagnostic performance. RESULTS: The analysis in this study is based on clinician-labeled LMD data obtained from the Vanderbilt University Medical Center (VUMC). We find that diagnosing ETV shows sensitivity to sustained phonation within the current database. Meanwhile, the results show that the proposed multi-stage transfer learning strategy can produce (1) accuracy of 65.3% on classifying normal and other three sub-disorders all at once, (2) accuracy of 85.3% in differentiating normal, ABSD, and ETV, and (3) accuracy of 77.7% for normal, ADSD and ETV. These findings demonstrate the effectiveness of the proposed approach.

Using Momentary Assessment and Machine Learning to Identify Barriers to Self-management in Type 1 Diabetes: Observational Study.

BACKGROUND: For adolescents living with type 1 diabetes (T1D), completion of multiple daily self-management tasks, such as monitoring blood glucose and administering insulin, can be challenging because of psychosocial and contextual barriers. These barriers are hard to assess accurately and specifically by using traditional retrospective recall. Ecological momentary assessment (EMA) uses mobile technologies to assess the contexts, subjective experiences, and psychosocial processes that surround self-management decision-making in daily life. However, the rich data generated via EMA have not been frequently examined in T1D or integrated with machine learning analytic approaches. OBJECTIVE: The goal of this study is to develop a machine learning algorithm to predict the risk of missed self-management in young adults with T1D. To achieve this goal, we train and compare a number of machine learning models through a learned filtering architecture to explore the extent to which EMA data were associated with the completion of two self-management behaviors: mealtime self-monitoring of blood glucose (SMBG) and insulin administration. METHODS: We analyzed data from a randomized controlled pilot study using machine learning-based filtering architecture to investigate whether novel information related to contextual, psychosocial, and time-related factors (ie, time of day) relate to self-management. We combined EMA-collected contextual and insulin variables via the MyDay mobile app with Bluetooth blood glucose data to construct machine learning classifiers that predicted the 2 self-management behaviors of interest. RESULTS: With 1231 day-level SMBG frequency counts for 45 participants, demographic variables and time-related variables were able to predict whether daily SMBG was below the clinical threshold of 4 times a day. Using the 1869 data points derived from app-based EMA data of 31 participants, our learned filtering architecture method was able to infer nonadherence events with high accuracy and precision. Although the recall score is low, there is high confidence that the nonadherence events identified by the model are truly nonadherent. CONCLUSIONS: Combining EMA data with machine learning methods showed promise in the relationship with risk for nonadherence. The next steps include collecting larger data sets that would more effectively power a classifier that can be deployed to infer individual behavior. Improvements in individual self-management insights, behavioral risk predictions, enhanced clinical decision-making, and just-in-time patient support in diabetes could result from this type of approach.

Integrating multiple blockchains to support distributed personal health records.

Blockchain technologies have evolved in recent years, as have the use of personal health record (PHR) data. Initially, only the financial domain benefited from Blockchain technologies. Due to efficient distribution format and data integrity security, however, these technologies have demonstrated potential in other areas, such as PHR data in the healthcare domain. Applying Blockchain to PHR data faces different challenges than applying it to financial transactions via crypto-currency. To propose and discuss an architectural model of a Blockchain platform named "OmniPHR Multi-Blockchain" to address key challenges associated with geographical distribution of PHR data. We analyzed the current literature to identify critical barriers faced when applying Blockchain technologies to distribute PHR data. We propose an architecture model and describe a prototype developed to evaluate and address these challenges. The OmniPHR Multi-Blockchain architecture yielded promising results for scenarios involving distributed PHR data. The project demonstrated a viable and beneficial alternative for processing geographically distributed PHR data with performance comparable with conventional methods. Blockchain's implementation tools have evolved, but the domain of healthcare still faces many challenges concerning distribution and interoperability. This study empirically demonstrates an alternative architecture that enables the distributed processing of PHR data via Blockchain technologies.

A mobile app identifies momentary psychosocial and contextual factors related to mealtime self-management in adolescents with type 1 diabetes.

Effective diabetes problem solving requires identification of risk factors for inadequate mealtime self-management. Ecological momentary assessment was used to enhance identification of factors hypothesized to impact self-management. Adolescents with type 1 diabetes participated in a feasibility trial for a mobile app called MyDay. Meals, mealtime insulin, self-monitored blood glucose, and psychosocial and contextual data were obtained for 30 days. Using 1472 assessments, mixed-effects between-subjects analyses showed that social context, location, and mealtime were associated with missed self-monitored blood glucose. Stress, energy, mood, and fatigue were associated with missed insulin. Within-subjects analyses indicated that all factors were associated with both self-management tasks. Intraclass correlations showed within-subjects accounted for the majority of variance. The ecological momentary assessment method provided specific targets for improving self-management problem solving, phenotyping, or integration within just-in-time adaptive interventions.

Analyzing the performance of a blockchain-based personal health record implementation.

BACKGROUND: The Personal Health Record (PHR) and Electronic Health Record (EHR) play a key role in more efficient access to health records by health professionals and patients. It is hard, however, to obtain a unified view of health data that is distributed across different health providers. In particular, health records are commonly scattered in multiple places and are not integrated. OBJECTIVE: This article presents the implementation and evaluation of a PHR model that integrates distributed health records using blockchain technology and the openEHR interoperability standard. We thus follow OmniPHR architecture model, which describes an infrastructure that supports the implementation of a distributed and interoperable PHR. METHODS: Our method involves implementing a prototype and then evaluating the integration and performance of medical records from different production databases. In addition to evaluating the unified view of records, our evaluation criteria also focused on non-functional performance requirements, such as response time, CPU usage, memory occupation, disk, and network usage. RESULTS: We evaluated our model implementation using the data set of more than 40 thousand adult patients anonymized from two hospital databases. We tested the distribution and reintegration of the data to compose a single view of health records. Moreover, we profiled the model by evaluating a scenario with 10 superpeers and thousands of competing sessions transacting operations on health records simultaneously, resulting in an average response time below 500 ms. The blockchain implemented in our prototype achieved 98% availability. CONCLUSION: Our performance results indicated that data distributed via a blockchain could be recovered with low average response time and high availability in the scenarios we tested. Our study also demonstrated how OmniPHR model implementation can integrate distributed data into a unified view of health records.

FHIRChain: Applying Blockchain to Securely and Scalably Share Clinical Data.

Secure and scalable data sharing is essential for collaborative clinical decision making. Conventional clinical data efforts are often siloed, however, which creates barriers to efficient information exchange and impedes effective treatment decision made for patients. This paper provides four contributions to the study of applying blockchain technology to clinical data sharing in the context of technical requirements defined in the "Shared Nationwide Interoperability Roadmap" from the Office of the National Coordinator for Health Information Technology (ONC). First, we analyze the ONC requirements and their implications for blockchain-based systems. Second, we present FHIRChain, which is a blockchain-based architecture designed to meet ONC requirements by encapsulating the HL7 Fast Healthcare Interoperability Resources (FHIR) standard for shared clinical data. Third, we demonstrate a FHIRChain-based decentralized app using digital health identities to authenticate participants in a case study of collaborative decision making for remote cancer care. Fourth, we highlight key lessons learned from our case study.

Mobile Momentary Assessment and Biobehavioral Feedback for Adolescents with Type 1 Diabetes: Feasibility and Engagement Patterns.

BACKGROUND: Integration of momentary contextual and psychosocial factors within self-management feedback may provide more specific, engaging, and personalized targets for problem solving. METHODS: Forty-four youth ages 13-19 with type 1 diabetes (T1D) were provided a Bluetooth meter and completed the 30-day protocol. Participants were randomized to "app + meter" or "meter-only" groups. App + meter participants completed mealtime and bedtime assessment each day. Assessments focused on psychosocial and contextual information relevant for self-management. Graphical feedback integrated self-monitored blood glucose (SMBG), insulin, and Bluetooth-transmitted blood glucose data with the psychosocial and contextual data. App + meter participants completed an interview to identify data patterns. RESULTS: The median number of momentary assessments per participant was 80.0 (range 32-120) with 2.60 per day. By 2 weeks participants had an average of 40.77 (SD 12.23) assessments. Dose-response analyses indicated that the number of app assessments submitted were significantly related to higher mean daily SMBG (r = -0.44, P < 0.05) and to lower% missed mealtime SMBG (r = -0.47, P < 0.01). Number of feedback viewing sessions was also significantly related to a lower% missed mealtime SMBG (r = -0.44, P < 0.05). Controlling for baseline variables, mixed-effects analyses did not indicate group × time differences in mean daily SMBG. Engagement analyses resulted in three trajectory groups distinguished by assessment frequencies and rates of decline. Engagement group membership was significantly related to gender, mean daily SMBG, and HbA1c values. CONCLUSIONS: Momentary assessment combined with device data provided a feasible means to provide novel personalized biobehavioral feedback for adolescents with T1D. A 2-week protocol provided sufficient data for self-management problem identification. In addition to feedback, more intensive intervention may need to be integrated for those patients with the lowest self-management at baseline.

A Smartphone App for Families With Preschool-Aged Children in a Public Nutrition Program: Prototype Development and Beta-Testing.

BACKGROUND: The Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) in the United States provides free supplemental food and nutrition education to low-income mothers and children under age 5 years. Childhood obesity prevalence is higher among preschool children in the WIC program compared to other children, and WIC improves dietary quality among low-income children. The Children Eating Well (CHEW) smartphone app was developed in English and Spanish for WIC-participating families with preschool-aged children as a home-based intervention to reinforce WIC nutrition education and help prevent childhood obesity. OBJECTIVE: This paper describes the development and beta-testing of the CHEW smartphone app. The objective of beta-testing was to test the CHEW app prototype with target users, focusing on usage, usability, and perceived barriers and benefits of the app. METHODS: The goals of the CHEW app were to make the WIC shopping experience easier, maximize WIC benefit redemption, and improve parent snack feeding practices. The CHEW app prototype consisted of WIC Shopping Tools, including a barcode scanner and calculator tools for the cash value voucher for purchasing fruits and vegetables, and nutrition education focused on healthy snacks and beverages, including a Yummy Snack Gallery and Healthy Snacking Tips. Mothers of 63 black and Hispanic WIC-participating children ages 2 to 4 years tested the CHEW app prototype for 3 months and completed follow-up interviews. RESULTS: Study participants testing the app for 3 months used the app on average once a week for approximately 4 and a half minutes per session, although substantial variation was observed. Usage of specific features averaged at 1 to 2 times per month for shopping-related activities and 2 to 4 times per month for the snack gallery. Mothers classified as users rated the app's WIC Shopping Tools relatively high on usability and benefits, although variation in scores and qualitative feedback highlighted several barriers that need to be addressed. The Yummy Snack Gallery and Healthy Snacking Tips scored higher on usability than benefits, suggesting that the nutrition education components may have been appealing but too limited in scope and exposure. Qualitative feedback from mothers classified as non-users pointed to several important barriers that could preclude some WIC participants from using the app at all. CONCLUSIONS: The prototype study successfully demonstrated the feasibility of using the CHEW app prototype with mothers of WIC-enrolled black and Hispanic preschool-aged children, with moderate levels of app usage and moderate to high usability and benefits. Future versions with enhanced shopping tools and expanded nutrition content should be implemented in WIC clinics to evaluate adoption and behavioral outcomes. This study adds to the growing body of research focused on the application of technology-based interventions in the WIC program to promote program retention and childhood obesity prevention.