You are here

Prototype Application of Mobile, Cloud-based, Watson-Like Technologies for TBI/PTSD Clinical Decision Support and Predictive Analytics


OBJECTIVE: Explore the use of natural language and clinical language understanding technologies, combined with IBM Watson-like technologies, to predict provisional diagnosis, provide clinical decision support, predictive analytics, and improved outcomes for mild TBI and PTSD patients. Develop a mobile, cloud-based architecture that can integrate with existing or improved clinical workflow, and ties to the Electronic Health Record. DESCRIPTION: For more than 30 years, academic researchers have studied the use of computers to predict diagnoses and make recommendations for treatment. Such work has developed a number of standards and frameworks for developing and executing clinical decision support guidelines and algorithms, such as SAGE, EON, GLIF, and ATHENA. Several commercial products to couple knowledge with problems, and to predict diagnosis and recommend treatments, such as PKC Couplers, Agile Diagnosis, Isabel, TheraDoc, Zynx, and others, and have emerged on the marketplace but have not been widely adopted, perhaps because the clinical decision support guidelines are typically segregated to one disease domain, and do not cross domains. In addition, such clinical decision support systems have not been truly integrated into a clinician"s workflow and/or into the Electronic Health Record. Maintaining current medical knowledge within the algorithms has also posed considerable challenge. The TATRC Morningside Initiative, started in 2007, was an attempt at a private-public partnership to create an open source repository of clinical knowledge. Two recent developments may change adoption rates, however. First, Qualcomm has joined forces with X-Prize foundation to over a $10M prize to any individual or company that can develop a mobile computerized decision support tool that can diagnosis patients as good as a panel of board certified physicians. Second, using technology that has evolved out of the widely advertised Watson Jeopardy game challenge, IBM has announced a major collaboration with WellPoint to use Watson technologies to diagnosis and recommend treatments for oncology patients. Several key partnerships have formed between Nuance, IBM, and 3M to apply speech recognition, natural language processing, and medical ontologies towards development of improved clinical decision support tools to predict diagnoses and offer treatment options. A number of companies are also releasing products that use embedded sensors to collect patient physiological signs, and include them as input to the clinical decision support algorithms. Some of these products are smart-phoned based, although research in mobile, cloud-based technologies is just beginning and is 5 to 10 years from maturation. PHASE I: In Phase I, the awardee will develop alternative strategic, operational, and technical architectural views for a clinical decision support aid which can predict diagnosis from patient history, symptoms, and/or physiological signs, with a focus on using mobile, cloud-based speech recognition, natural language processing, and IBM Watson-like technologies to capture and analyze data. Initial use case will focus on improving outcomes for mild TBI and PTSD patients. The awardees will work with TATRC and its partners to incorporate existing, ongoing natural language processing and clinical decision support work as applicable, and produce a complete design document for a clinical decision support tool that will predict diagnoses based on patient history and symptoms. The government encourages evaluating the use of cloud-based algorithms which can be executed on mobile devices. The government also encourages the use of standards based, open source clinical practice guidelines for execution. To the extent that the specific pathways can be seen within the algorithm, and are not proprietary, they can also be used to support training of residents. PHASE II: In Phase II, one or more awardees from Phase I will build and test a prototype. TATRC can provide use of a virtualized Early Stage Development (ESP) platform that can be used in the project, although a vendor use fee will be negotiated under a CRADA. The ESP platform has access to a fully functional AHLTA and CHCS development and test environment. TATRC will also make available open source VISTA code that the awardees can consider using in the project. TATRC will attempt to supply sufficient military clinical subject matter expertise to test and evaluate the prototype as to clinical relevance, accuracy, and usability. The awardees should also provide their own clinical expertise in addition to what expertise the government may provide. PHASE III: It is anticipated that this research will yield improvements in application of mobile-based natural language processing, clinical language understanding, text analytics, computerized algorithms, and artificial and business intelligence, that may have applicability outside of the medical domain and may be applied in other domains, such as military intelligence. The government also expects that this research may yield new developments in computerized modeling and simulation to train personnel in a number of industries.
US Flag An Official Website of the United States Government