Initial Implementation of an Intelligent Patient Data Review Assistant (IPDRA) LTC Trinka Coster, MD Matthew Medlock, MD LTC Joseph Parker, MD Jim Ong,

Initial Implementation of an Intelligent Patient Data Review Assistant (IPDRA)

LTC Trinka Coster, MDMatthew Medlock, MD

LTC Joseph Parker, MDJim Ong, MS

Jane Dowling, MSStephen Porter, MDSeth Powsner, MD

Ida Sim, MD PhD

Walter Reed Army Institute of ResearchUniformed Services Univ. of the Health SciencesWalter Reed Army Medical CenterStottler HenkeNYU Medical CenterHarvard Medical SchoolYale University School of MedicineUC San Francisco School of Medicine

ATA 2004

TATRC

PROBLEM• Automated clinical results require users

to shift through voluminous, un-summarized electronic data which are not formatted in a way to help the physician discover and/or see relationships

• No method to communicate to and from a medical consultant integrated medical data to facilitate diagnosis or management of a medical problem

TATRC

DIA – Data Standards for Clinical Development – Sept. 17, 2002

Patient Profiles and Standard Submission Data

Graphical Patient Profile

Small Business Innovative Research Project – Cognitive

Patient-Clinician Encounter Model

.

Mission

Project Goals

Testbed

Integrate patient data to facilitate rapid identification of trends, interactions, and/or medically relevant relationships

Develop and evaluate an operational prototype that demonstrates feasibility and utility

Dept. of Defense clinical databases

Intelligent Patient Data Review Assistant Strategy (IPDRA)

Patient data views

Information-dense displays

Web access

Visual view authoring

Integrate data across disparate sources to present clinically-meaningful subsets of medical data by problem, guideline or concept

MultiTimeGraphs displays coordinated timelines and time-series graphs

Web browser displays views generated by the IPDRA web application server

Users specify view logic by using authoring tool to draw flow charts

Research Questions

Utility of views

View design principles

Feasibility

Do views decrease the time for patient data review by clinicians? Do views make it easier to detect trends and/or relationships? Do views decrease medical errors?

What design guidelines and principles can help ensure effective views?

Does the authoring tool enable practical and economical creation of potentially large libraries of views?

Cardiac View (ICDB)

Hypertension View (ICDB)

Diabetes View / summary (ICDB)

MultiTimeGraphs mockup

MultiTimeGraphs mockup

IPDRA Architecture

View GenerationView Authoring

Authoring Tool (SimBionic)

View logic specifications

IPDRA Web Application Server or IPDRA View Test/Debug Application

SimBionic Run-time System

predicates: db utils

actions: HTML, XML generation

clinical database(ICDB, M2, CHCS II)

IPDRA database (graph, SQL, HTML

templates)

IPDRA View Authoring Tool

Preliminary View Design Findings• Data review steps can be inferred from

clinical practice guidelines to guide view design

• For complex problem views, physician can use authoring tool to visualize integrated data

• Views can show physical exam results, risk factors, related problems (ICD9), labs, pharmacy, procedures and radiology

TATRC

IPDRA Status• Designed and implemented prototype of IPDRA

authoring and view generation software• Prototyped 3 views (hypertension, cardiac

overview, diabetes mellitus) using DoD’s Integrated Clinical Data Base (ICDB) as the test database

• Evaluation (and refinement) of views planned for fall 2004 IF WE GET ACCESS TO SANITIZED CLINICAL DATA

TATRC

ISSUES

• Access To Sanitized Clinical Data• Electronic Drug Label – Knowledge Source• Electronic Guidelines – Knowledge Source• Redefine “Drug Allergy Section” to “Allergy &

Drug Intolerance and Ineffectiveness”• Medical Confounders: “Yes/No” “Start Date/Stop

Date”• Edit and Annotation Features

TATRC