![]() ![]() ![]() The VDW was initiated during the 1990s with support from the Cancer Research Network and increasingly expanded and standardized over the years with support from many other federally-sponsored research networks. PRISM builds on many years of development and research use of the Virtual Data Warehouse (VDW), a data model developed by the Health Care Systems Research Network to facilitate public domain health and health services research. PRISM generates personalized prognostic information by gathering key characteristics about the index patient, identifying a cohort of similar patients that are matched to the index patient on these characteristics, and displaying the survival curves of the similar patients. In response, we created a prototype Prognostic Information System (PRISM) to enable physicians to generate personalized information for individual patients during real-time clinical encounters. In 2015, a KPNC oncology leader (D.M.B.) requested that our organization’s Division of Research develop a tool to generate personalized information from our own health care databases for use in counseling cancer patients. Like all Kaiser Permanente regions, KPNC uses EPIC as the core of its EMR. More than 17,000 patients in this population are newly diagnosed with cancer each year. Our integrated health care system, Kaiser Permanente Northern California (KPNC), includes 21 medical centers and more than 100 oncologists. This project was conducted by a multidisciplinary team within The Permanente Medical Group, which partners with Kaiser Foundation Health Plan to serve more than 4 million patients in Northern California. This case study describes how we developed a prototype that overcomes these barriers to provide physicians with real-time prognostic information tailored to individual patients for use during clinical encounters. A second major barrier is the technologic complexity of using electronic medical record data to create predictions rapidly enough to provide information to individual patients during clinical encounters. A major barrier has been the lack of comprehensive patient-level data over multi-year periods that cover the course of the patient’s disease trajectory and long-term outcomes including survival. In parallel, physicians would value tools that can rapidly analyze and display outcomes in the course of their conversations with patients.Īttempts to create “patients like me” tools by organizations such as the MD Anderson Cancer Center and the American Society of Clinical Oncology have not yet succeeded despite large financial investments. The need for personalized information has led to the development of “patients like me” websites and tools that connect groups of patients based on key characteristics. Providing data on the actual outcomes of similar patients would enable conversations that are more personalized and credible than those based solely on published literature. In oncology and other specialties, physicians could benefit from a tool that rapidly identifies and analyzes a group of patients similar to the patient they are counseling. Oncologists are often asked by patients newly diagnosed with cancer, “Have you ever seen a patient like me?” This question may convey the patient’s need to be treated as an individual as well as concern about the oncologist’s experience with their particular cancer, especially in the current era of targeted therapy. However, a gap remains between the promise of data science to transform health care and the limitations of implementing reliable user-tested tools in actual practice. More than 9 of 10 physician practices have EMRs, creating the potential to use data to inform clinical decisions and support learning health systems. Physicians and health systems are eager to reap the benefits of using the evidence that can be generated from electronic medical records (EMRs), yet are frustrated that more useful tools are not yet available.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |