background: The California Department of Public Health (CDPH) data collection systems provide essential information necessary for policy implementation to improve the health of all Californians. Unfortunately, the current culture of siloed data collection systems inhibits the ability to enhance, exchange, and integrate valuable information. The Ecosystem of Data Sharing (EODS) Interoperability Initiative is novel technology being implemented at CDPH to create a common integrated platform for public health information.

methods: This technology solution adapts the agile approach, while adhering to legislative mandates, semantic, technical, and functional interoperability standards; and utilizes use case development strategies to incrementally build a data hub framework infrastructure. EODS employs a semantic graph database to connect and build relationships within the data from three registries- the patient, facility, and provider. These registries serve as the core reference indices to identify and merge duplicate records of individuals at ingestion. Current data sources integrated include: CDPH Vital Records, The California Office of Statewide Health Planning and Development, The California Parkinson’s Disease Registry, California Cannabis Surveillance, and California Emergency Medical Services Authority.

results: Integration of these data sources allows EODS to provide Data as a Service to over 200 programs within CDPH. This creates a robust patient record using probabilistic algorithms, while simultaneously maintaining data provenance, operationalizes this technology and enables CDPH to move towards automated data mining and machine learning capabilities that will drive predictive analytics.

conclusion: Implementing this novel technology within CDPH, improves data robustness by providing integrated data and data linkage. Further advances to the EODS infrastructure will establish EODS as model surveillance system. Enabling CDPH to conduct real time surveillance and foster data driven decisions that will improve the overall health outcomes of Californians.

Background

In recent years, researchers have turned the text of the public health laws into numeric legal data through policy surveillance. This legal data can be analyzed alongside outcome data to study the impact of legal interventions on public health. To analyze legal data alongside outcome data, researchers must download, clean, merge, and analyze data sources to disseminate findings in discrete operations using multiple software tools.

Objective

Offering a new policy analysis paradigm, our team utilized the power of our policy tracking software to create legal data on opioid-related laws like naloxone access laws and medication assisted treatment laws. We wanted to create a platform where this legal data can be uploaded alongside outcome data to help streamline and accelerate the assessment of policy impacts through a dynamic policy analysis laboratory.

Methods and Results

Using human-centered design principles, cloud storage, distributed cloud architecture and containerized virtual machine technologies, we built a policy and outcome evaluation platform. We published this platform to serve as a researcher-friendly website that facilitates the overlay of health outcome data on policy data, as well as integrated computational functionality in major statistical programing languages for further analysis. We worked with an external group to develop an analytical model using our opioid-related policy data and existing outcome data to assess the impact of key state legal interventions on reducing the opioid epidemic.

Conclusions

We developed the policy and outcome evaluation platform as a web-based laboratory for health policy and outcome data, allowing researchers to launch and share policy evaluations with ease. By adding in source credibility features and streamlining data resource sharing, we can now leverage the platform to be an online community of data for researchers to explore, analyze and publish datasets or codes in a data-science environment centered on health policy outcomes.

Background: Telehealth is increasingly used to improve healthcare access, cost, experience, and effectiveness. Despite its many advantages, telehealth use in oncology is uncommon nationally and provider perceptions regarding the benefits and barriers to its uptake are not well understood. We completed a qualitative study evaluating oncology providers’ perceptions of video visits at an institution with a system-wide telehealth program.

Methods: We performed semi-structured interviews with oncology providers at Thomas Jefferson University to assess current use, perceived barriers to use, and perceptions of telehealth impact on access, cost, experience, and effectiveness. All interviews were transcribed and coded using a team-developed codebook, with frequent assessments to ensure interrater-reliability across codes. Codes were grouped into themes and subthemes and organized into a conceptual model of barriers to telehealth.

Results: Twenty-eight providers (20 oncologists and 8 advanced practice providers) completed interviews, twenty-five of whom utilized video visits. The majority of providers viewed video visits favorably, with key perceived benefits including immediacy of the visits, increased patient convenience and interactions, and the option to have multi-person visits. Key perceived barriers included patient and provider unfamiliarity with telehealth, patient privacy and cost concerns, work routine disruption, regulatory and technology challenges, and limitations of video visits compared to office visits (e.g. physical exam and emotional connection).

Conclusions: Telehealth is generally viewed favorably by oncology providers within a system with widespread telehealth. There are several perceived barriers, most pertaining to logistics and knowledge limitations, which should be addressed to support increased telehealth uptake.

Background: The prevalence of Alzheimer’s disease is higher in Hispanics and African Americans than non-Hispanic Whites. Culturally appropriate interventions are crucial to support Hispanic and African American family caregivers for persons living with Alzheimer’s disease and related dementias (ADRD). Deep learning has potential to assist personalized recommendations based on their contextual, continuous lifestyle habits and situational factors among Hispanic and Black family caregivers of ADRD. We visualized topics and its semantic relationships from health literature that use deep learning, exploring the techniques’ ability to provide a foundation for developing culturally appropriate data science-based interventions for Hispanic and African American dementia caregivers.

Methods: We extracted unstructured text from titles and abstracts of 8,060 studies mentioning deep learning from PubMed library. We applied natural language processing (NLP) and the Girvan-Newman clustering algorithm to detect topics and their semantic relations in the corpus. Topics and their relations were visualized using a network diagram.

Results: Literatures applying deep learning in health science (N=8,060 studies) are limited to five clusters including 1) modeling approach, 2) human disease diagnosis, 3) clinical datasets, 4) neural network, and 5) drug discovery and genetics (Newman modularity value = 0.44). ADRD and drug discovery were detected as emerging areas for adapting deep learning in health.

Conclusion: Although deep learning is now common in art and science research as well as industrial applications and pervasive in our daily lives, its use in health science remains uncommon, with a slow rate of adoption. While these uses are largely concentrated on limited topics such as disease diagnostics and biology, topics related to behavioral science were rarely found. Applying deep learning methods on various cultural artifacts may provide insights for developing culturally appropriate technology-based interventions for family caregivers of persons with ADRD.

background

Spinal cord injury is a condition that presents in many forms – complete, incomplete, chronic, and acute. People with spinal cord injury experience many changes in their quality of life. One of the main adjustments is learning to cope with disability and manage their pain. Opioids are heavily prescribed to this population, and differences in opioid addiction exist between patients with spinal cord injury and patients with other common surgeries and health conditions.

objective/purpose

The aim of this project is to evaluate differences in opioid addiction between patients with spinal cord injury and patients with other common health conditions; determine if racial differences exist in opioid addiction in patients with spinal cord injury; and assess health disparities in patients with spinal cord injury. The focus will be on patients in the Epic medical record in Saint Louis, the electronic health record system for Barnes-Jewish Hospital and St. Louis Children’s Hospital (BJC), Washington University School of Medicine, and other local institutions.

methods

Self-service queries using the Epic SlicerDicer application allow the research team to include patient diagnoses of interest and appropriate inclusion and exclusion criteria before downloading a CSV file of millions of patients. This project uses a machine learning approach to predict opioid addiction, as well as multilevel modeling in R. Patients are nested within zip codes.

results

Preliminary results show that there are higher rates of opioid addiction in spinal cord injury patients when compared to patients in the general population. Furthermore, there are differences in the opioid addiction rate between black and white spinal cord injury patients as black spinal cord injury patients have a higher rate of addiction than whites.

discussion/conclusions

Further research will evaluate the relationship between socioeconomic status and access to physical therapy services since physical therapy has been associated with lower opioid use.

This presentation describes and analyzes a collaborative process within a major engineering professional society addressing trustworthiness of interoperable health informations systems. The collaboration seeks to (1) reduce risk of violence through health informations systems sabotage, misuse, or erroneous use and (2) proactively build privacy-protections, security, and trustworthiness (for patients, clinicians, systems, and researchers) into ongoing efforts to improve and standardize health IT interoperability.

The presentation assesses risks to violence through health IT misuse, explaining how the interoperability of information systems so desirable for the creation of iterative health learning systems increases that risk.

It briefly sketches the rationale and history of a major response within an engineering professional organization: the IEEE Data and Device Interoperability with TIPPS Working Group. (TIPPS =Trust, Identity, Privacy, Protection, and Security.) It sketches how subgroup organization integrates technical, ethical, and regulatory challenges. For example there subgroups on Trust and Identity, Privacy, Security, and Protection and Safety work together with subgroups on Data Validation and Interoperability, Device Validation and Interoperability, and Intelligent Systems Design.

It then provides two illuminative examples.

The first example explains how one key ethical question faced by the working group is intertwined with one key engineering challenge.

The second example presents one case study being developed by the working group. (The working group's set of case studies includes "bad scary" and "good proactive" cases. Final choice of which case study will be presented will be made as that case-study set is filled out and refined.)

Background: The growing use of the internet among Latinos to seek health information offers an opportunity for the wide dissemination of evidence-based resources. The objective of this project was to increase access to evidence-based health information among Spanish speaking Latinos through a podcast and social media video content.

Methods: A multidisciplinary team of health professionals, researchers, and librarians researched four areas focused on the Latino population, including most common health-related online searches, optimal platforms, dissemination strategies, and prevalent health conditions. Then, worked with a community advisory board to develop the format of the online content and pilot-tested a beta version with lay community members. The team also identified bilingual health professionals to interview in the episodes and developed an airing calendar. Finally, the team worked with a podcast coach for eight weeks to integrate all components of the program and received guidance from social media experts.

Results: The team recorded 20 twenty-minute audio and video episodes relevant to the Latino community’s health needs and culture. The episodes are appropriate for dissemination via social media (i.e., Facebook, WhatsApp, YouTube) and podcast (i.e., Pandora, Spotify, SoundCloud) platforms. Moreover, they developed a start-up system that can be used by others interested in podcasts and social media to disseminate health information.

Conclusion: Podcasts and social media can be used to share health information that is culturally and linguistically tailored to Spanish-speaking Latino populations. Our next step includes assessing the effectiveness of this program in terms of reach, knowledge, and use of resources, and health behaviors.

[Objective]

Clinical pharmacy care combines NHI-PharmaCloud, and integrates the drug safety assessment process, achieves the integration of medical care manpower, and effectively implements clinical pharmacy care related to drug safety.

[Methods]

This study targeted 159 patients in a psychiatric day ward in a regional teaching hospital in the north from February 2018 to June 2019. The pharmacists performed PDCA cycle management for psychiatric patients' medications. The team uses creative services to continuously improve the quality of care. The pharmacist will perform a descriptive statistical analysis of the evaluation records to understand and analyze the monitoring of the quality of medications for mental patients.

[Results]

After research and analysis, the focus is on the following key patients, including ICD-10 is F20~F25 series, patients with schizophrenia or schizoaffective disorder, and comorbidities of metabolic diseases. It’s according to the medication safety assessment, the medication care model of visualized red, yellow, and green light medication safety grade information is established. The sticker is affixed to the "Medicine Care Consultation Sheet for Inpatients in Psychiatry" to increase the recognition and reminder function.

Through the pharmaceutical care service and cross-team discussions, doctors' prescription behavior has changed, the safety assessment level has reduced the high alert of the red light is 75%, and increased the relative safety information of the green light is 34%.

[Conclusions]

The hospital launched a clinical interprofessional cooperative care mode for the psychiatric day care unit. The physician appropriately adjusted the medicine and its dosage to reduce the incidence of comorbidities. The pharmacist provides appropriate drug information and clinical pharmacy services. The nurse provides a bridge between the patient and the medical team, encourages patients to participate in the entire treatment plan, and strives to cooperate to implement the best treatment.

Background: Indian women face the largest digital divide in the region. However, our formative research points out that young men and women have similar levels of phone use in Odisha, India. In fact, women surpass men in daily (83.30%) and smartphone use (10.0%). The focus of this research effort is to assess differences in mobile phone use, demographics, and digital literacy levels among Oriya women.

Methods: In 2019, we used a random sampling method to collect data on indicators of mobile phone use, digital literacy, and demographics data from 1874 women of reproductive age (15-49 years old), in Odisha, India. Descriptive and Analysis of Variance (ANOVA) with Tukey’s post-hoc tests assessed differences among groups delineated according to their mobile phone ownership.

Results: Women in Odisha, India are either exclusive mobile phone owners (43.97%), mobile phone borrowers who require permission always or most of the times prior to use (42.63%), or non-users (13.39%). Exclusive phone owners are the youngest, most educated and digitally literate, and have fewer children. Majority of exclusive phone owners are married (80.34%) and do not belong to a scheduled tribe (78.40%). Phone borrowers trail second behind exclusive owners on all measures followed by non-users. Significant group differences emerged in education and digital literacy between all three groups. Majority of phone owners also share their phones (58.38%). Phone sharing and borrowing relationships are primarily limited to male partners, children, and female relatives.

Conclusion/Discussion: There are three distinct groups of mobile phone users in Odisha with significantly different education and digital literacy levels. These groups require different intervention strategies to maximize reach, exposure, and engagement. mHealth interventions appear more tenable for exclusive owners while hybrid interpersonal and digital approaches may be more suitable for phone borrowers. Future mHealth interventions to be cognizant of these different groups when developing their target audience and strategies.

In order to develop strategies for intervention and prevention, knowledge about the specifics of violent deaths are necessary, including understandings of at-risk populations; risk factors, precipitating events, and circumstances for homicide and suicide; weapons or methods used in violent death; the role of drugs and alcohol in violent death; and communities and localities with high numbers or rates of violent death, the Office of the Chief Medical Examiner (OCME) in the Virginia Department of Health (VDH) has been participating in the NVDRS since it was established in the Centers for Disease Control and Prevention (CDC) in 2002. The unique nature of being housed in the Office of the Chief Medical Examiner, allows coders have access to a host of data sources, including medical examiner reports, law enforcement reports, forensic and toxicology reports, medical files, and a variety of community media outlets. These data sources are synthesized to build a more complete picture of the events leading to and occurring after the death of an individual. The National Violent Death Reporting System (NVDRS) is a unique tool designed to assist states in answering these questions about the who, what, when, where, and why of violent death. The VVDRS has grown to become the go-to program for information about violent death in Virginia. Because VVDRS links demographic data with circumstance, toxicology, location of injury and of death, and mechanism of injury, prevention and planning stakeholders in Virginia have come to rely on VVDRS to understand and respond to violence in their communities.