The technical innovations of the proposed project are: the design, implementation, and evaluation of a telemedicine-based approach for improving access to care and for collecting data on health disparities in underserved populations. This is a new telemedicine framework that provides the functionality required for Urban Telemedicine. Equally important, the proposed framework includes the high reliability and performance that is required in order to operate a telemedicine system in the proposed environment. The framework is scalable to multiple sites (each with their own data models and schemas), and enables researchers to incorporate culturally-sensitive surveys. The framework is highly portable to multiple platforms, and is freely available to other researchers so that they can test and validate the results obtained in the proposed Urban Telemedicine environment in other urban or rural settings.
The clinical research impact of the proposed project is that it will provided a needed test-bed to test the hypothesis that such telemedicine systems are needed to improve access to care. In the course of implementation, the proposed system will also improve the collection of health information from disadvantaged and poorly-studied populations, and it will increase access to health information for studies of disadvantaged populations. Longer-term, it will decrease the cost of treating disadvantaged populations by identifying problems earlier, and it will allow comparisons across populations in different geographic locations and diverse environmental conditions. This may lead to badly needed improvements in health outcomes in the target populations.
Over the past few years the Research Informatics Core developed a variety of telemedicine and teleinformatics capabilities, including telemedicine data collection and display software, a multi-institutional shared brain imaging database and documentation and policies for interaction between collaborating locations. Core members collaborated to publish numerous papers and invited presentations in these areas, some of which are summarized here.
Research Informatics Core members have extensive experience in the design and development of telemedicine systems. The scope of this expertise encompasses the design and optimization of networks for Teleradiology [Duerinckx, Valentino et al, 1996], the development of Internet-based telemedicine systems [So, Li et al, 1996; Li, Valentino et al 1996], the development of multimedia patient-records management systems [Valentino, Huang et al, 1998], and the research and development of advanced tele-monitoring for scanners and operating rooms [Ratib et al, 2000; Farahani, Valentino et al, 1999].
Research and development has focused on the required teleinformatics resources, including hierarchical storage servers for the distributed archive and retrieval of medical imaging data [Huang, et al, 1998; Valentino, Lufkin, et al, 2001; Valentino, Wei, Flowers, et al, 2002]. Recently a multi-tier storage infrastructure was implemented in Java using JSP, J2EE, JDBC and mySQL for the internal relational database. The underlying technology has been successfully used in a variety of imaging research projects, and this technology will be used for the storage and distribution activities within this project.
DATA MODELS AND SCHEMA INTERPRETATION
Recently, approaches were investigated for mapping data elements and semantics from one database to a desired output file or other database. As a result, a graphical data translation engine was designed and developed that enables reading and writing from various data sources, interpretation of the data elements, and translation of the attributes from one data model to another [Valentino, 2003]. This approach enables the use of graphical tools to quickly and easily select data elements, specify how to interpret them, and map them to an output data source.
The research team also has extensive experience in the design and development of clinical workstations and presentation systems [Harreld, Valentino, et al 1998; Valentino, Harreld, et al 1998]. Recently team members designed and developed a portable framework in Java (the jViewbox/Tk) for the presentation of brain imaging data [Valentino, Ma et al, 2002]. The framework provides a comprehensive set of Java classes and an easy-to-use application programmer interface and programming model to simplify building imaging-intensive medical applications. The jViewbox was evaluated on a variety of platforms and optimized to provide excellent performance for very large data sets. jViewbox was also used to build graphical user interfaces on mobile devices and will be used to support the display and presentation of data in this project.
PROTOTYPE TELEMEDICINE SYSTEM
Over the past year the new, Open Source Urban Telemedicine System (UTS) was designed and implemented in a pediatric primary care setting. It currently enables the integration and management of clinical and research data through a centralized data center. Included are automated business processes to secure data exchange, tools for data collection and visualization, data screening, and a web based statistical analysis environment. These automated web services operate in a secure environment that extends across all departmental and organizational boundaries providing an effective means for real-time web based collaboration.
The principals on this project have extensive software engineering expertise. Team members have used software engineering procedures in their research work for many years [Valentino, Taira, et al, 1994; Valentino, 1996; Valentino, 1998], and have standardized software development using widely available and portable software technologies. This includes the use of Java and J2EE for toolkit and application development, JDBC for database access, and MySQL for relational database management. The team also developed a template to capture Functional Requirements for infrastructure software, and used the Unified Modeling Language to describe the functionality and "look and feel" of application user interfaces [Valentino, Fiske, et al, 2001]. All newly developed software is routinely tested for reliability and performance across a variety of operating systems to ensure the portability and robustness of software development efforts.
OTHER KEY INFRASTRUCTURE
Lastly, as the reviewers are aware, there are a number of other key infrastructure issues that cannot be described in detail in this proposal. However, as described in a number of related publications [Valentino et al, 1996; Li, Valentino et al, 1996; Valentino, Huang et al, 1998], the research team has developed not only the software resources, but also the procedures and strategies to address each issue. For example, security is a concern from both technical and policy standpoints. As described throughout this proposal, firewalls and authorization access controls, as well as the policies and procedures, have been implemented to ensure the security and confidentiality of subject data and other resources.
RATIONALE FOR TELEMEDICINE AS A TOOL FOR ADDRESSING HEALTH DISPARITIES
As described in the introduction to this proposal, numerous factors contribute to the significant health disparities that have persisted, and in some cases worsened, over the past decade between urban Caucasian populations and socially marginalized, economically disadvantaged groups. However, due to a lack of relevant clinical data, it is currently difficult to fully understand the extent of health disparities and their underlying causes. Furthermore, the number of sub-specialist physicians is decreasing nationwide. This is particularly acute in the underserved communities where the worst health disparities exist and sub-specialist expertise is perhaps most needed.
Two significant barriers to reducing health disparities include the difficulty of collecting, managing and sharing clinical research data (leading to a lack of relevant clinical data), and the difficulty of designing and developing the appropriate systems for evaluating new approaches to care (leading to a lack of effective systems for providing needed care). The focus of this research project is to design a new telemedicine system that addresses critical needs of underserved populations, and to develop an appropriate software framework that improves access to care and improves data collection for patients in underserved communities. The resulting software will be used to evaluate the ability of the proposed telemedicine approaches to reduce health disparities.
SOFTWARE DESIGN AND ARCHITECTURE
The open-source telemedicine software currently in use is highly reliable, secure, easy to maintain and easy for other researchers to adapt and extend as requirements change. Although some software exists to manage electronic medical records, little such software exists for imaging-based telemedicine. Over the past year, a number of design criteria and key system components have been documented, developed and tested. A flexible database schema design was proposed and developed [Valentino, Lufkin, et al, 2001] to enable the management of multiple data models. This will enable adaptation of the system to new clinics and new healthcare systems as they are encountered. An object-oriented design and development approach, which has been used on past projects, will ensure that the software is scalable and reliable [Harreld, et al, 1998; Valentino, et al, 1998]. Lastly, the inclusion of appropriate functionality will make the software easily configured for new health disparities research studies, and will facilitate compliance with the Healthcare Insurance Portability and Accountability Act (HIPAA).
A number of features of the system have already been designed and validated. As mentioned in Preliminary Results, a prototype system has already been deployed to monitor Diabetic Retinopathy at several community-based clinics in the South-Central Los Angeles area. Some of the functionality of the system is briefly described here to indicate the functionality that is included in the Pediatric Telemedicine system.
Secure login screens authenticate all users and restrict access to the appropriate medical records. Data Entry screens enable Physician Assistants and Front Office staff to enter and validate patient demographics either before the initial patient visit, or during the patient examination, as required. Clinical Care and Research Data Collection screens to will be implemented to collect clinical data for telemedicine care as well as research data for specific clinical research studies (with appropriate IRB approval).
The software is written in the Java language and Java Server Pages (JSP). It uses the Apache Web Server and Tomcat JSP Server, and the database uses MySQL. All of these components are free, open systems that are widely available to the research community. The major system components include: (a) the web-based, distributed and redundant Telemedicine Server that includes a database and permanent archival of clinical and research data collected; (b) the Exam Room Module that is used by the PA to collect clinical and research data; (c) the Reception Module that is used by the PA or Receptionist to register the patient and collect basic demographic information; and (d) the Evaluation Module that is used by the Specialist to review patient records and determine how to care for the patient.
Critical clinical data is captured using the Exam Room module. As illustrated in the Figure, as clinical data is captured, it is queued on the local system, and then sent over a secure and encrypted interface to the web-server. Only after the web-server confirms that the data was accurately received and archived, then the local system can delete that data.
The Reception and Evaluation modules run in a standard web-browser on any desktop computer. This makes it inexpensive to implement and maintain, and very convenient for the Specialist to access medical records as needed. The web-server, of course, requires more computer memory and significantly more disk space than the Reception and Evaluation modules. The only module that requires special hardware is the Exam Room module. If the exam images, then a PC image capture board is required to digitize and store high-resolution images from an imaging device such as a dermatology camera or otoscope. The use of USB-based image capture devices was evaluated, but such devices were too slow to capture high-resolution images in real-time. However, with the development of the new USB 2.0 standard, it is expected that such devices will be widely available within the next year.