Overview

In healthcare settings, healthcare consumers and providers both need quick and easy access to a wide range of online resources. Patients and their family members need information that can educate them about their personal situation, while clinicians, and particularly those still within the formal educational process, need information that is clinically relevant to an individual patient under supervision. Providing easy and focused access to those portions of medical knowledge that are needed to care for a specific patient can have enormous benefits. When the latest medical information is provided at the point of patient care, it can help practicing clinicians and physicians in training to avoid missed diagnoses, choose only effective interventions and diagnostic tests, and minimize impending complications. The latest medical information, when expressed in understandable terms, can empower patients to take charge of their healthcare, helping them to change lifestyles, take appropriate preventive measures, and make more informed choices regarding their treatment. It can improve the overall quality of healthcare, resulting in better outcomes because informed patients are more likely to comply with treatment plans and become a part of the healthcare team.

The goal of this proposal is to provide personalized access to a distributed patient care digital library through the development of a system, PERSIVAL (PErsonalized Retrieval and Summarization of Image, Video And Language resources). PERSIVAL tailors search, presentation, and summarization of online medical literature and consumer health information to the end user, whether patient or healthcare provider. PERSIVAL utilizes the online patient records available at Columbia Presbyterian Medical Center (CPMC) as a sophisticated, pre-existing user model that can aid in predicting user's information needs and interests. For those patients with no CPMC patient record, PERSIVAL asks specific questions, depending upon the user query and clinical context, to build and maintain a skeletal user health information model.

Key features of the proposed work include personalized access to distributed, multimedia resources available both locally and over the Internet, fusion of repetitive information and identification of conflicting information from multiple relevant sources, and presentation of information in concise multimedia summaries that cross-link images, video, and text. Given the widely varying nature of online resources, research in retrieval and search methodology focuses on automatically identifying source type (e.g., journal articles vs.\ self-help groups), quality, and level of intended audience. In addition to fusing information from multiple sources, summaries must also express facts in terms the user can understand, regardless of background. Video sources range from diagnostic test results to educational video, each of which requires search based on image characteristics and identification of significant events to aid in finding appropriate clips.

Research includes development of system components for:

query specification through a combination of input modalities along with facilities to automatically augment an input query with information from the patient records;
constrained search using information from the patient record over distributed resources that have been automatically indexed and classified according to source type, quality, and reading level;
presentation and summarization of multiple relevant documents, merging repetitive information and using language generation techniques to select everyday language when needed;
search and presentation of multimedia resources, supporting browsing over automatically constructed categories, techniques for automated summarization of video, and graphical layout and annotation of diverse material.

PERSIVAL is evaluated and tested in clinical settings within both the Heart Failure Center and Diabetes Center at CPMC. In addition, we plan to test our system with outpatients in the Milstein Patient Library and home settings. We carryed out formative evaluation in early stages using cognitive design and usability engineering to identify, for example, features needed in the user interface and the kind of information that should be provided in a summary. Summative evaluation in the final years of the project will quantify the impact of our system on physician and patient behavior and satisfaction. In order to facilitate a robust, operational system that can be deployed and maintained both within and beyond the period of the grant, we will rely on and extend security and authentication protocols developed at Columbia and will exploit the computerized infrastructure in place at CPMC to aid in integrating and sustaining PERSIVAL within the clinical environment.