# Improving Electronic Health Record Usability and Usefulness with a Patient-Specific Clinical Knowledge Base

> **NIH NIH R21** · UNIVERSITY OF ALABAMA AT BIRMINGHAM · 2022 · $161,494

## Abstract

Electronic health records (EHRs) are providing opportunities to revolutionize health care. However, they have
brought with them a number of burdens – some expected and others unanticipated. The medical literature is
replete with complaints about how important information in patient records is difficult to find, partly due to its
absence and partly due to its obfuscation by a proliferation of low-value data in what is called “note bloat”.
Other complaints focus on clinical alerting applications, which have proven to issue vastly more false alarms
than true ones, leading to alert fatigue which results in clinicians missing the important warnings. Reuse of
EHR data for research is also difficult. At this writing, multiple groups (ACT, eMERGE, All of Us, N3C and
others) are working to automatically identify patients with COVID-19 (SARS Var-2 infection phenotype) using
EHR data – a task that should be trivial, but clearly is not due to suboptimal EHR content and organization.
Extensive effort to data has not succeeded in resolving these complaints about EHRs.
The premise of the proposed work is that there is information about the clinicians’ thinking that is not readily
available or is missing from the EHR and that if it can be added in a structured, computable way EHR
improvements can follow. We refer to that information as the “why” of health care: why does the clinician think
the patient has a sign or symptom, why is a particular test or treatment being chosen, why is a treatment being
discontinued. The proposed work will explore way to represent patient data with this added knowledge to better
understand what additional information must be added to the EHR, how the addition might be accomplished,
and how the resulting knowledge base might be used. As a first step in usage, we will explore a knowledge-
based method for improving the navigation of patient data in an EHR.
The project will involve three sequential steps. First, we develop methods to break down the information in a
patient record, including information from narrative text (notes), into individual medical entities (such as
problems, tests and medications) to create patient data sets (PDSs). Next, we will build on our preliminary
studies of the concepts of the clinical care context (patient findings and conditions, diagnostic tests and their
results, and therapeutic plans) to add relationships between these entities that convey the clinical reasoning
behind them (such as linking a problem to set of possible causes, a test intended to differentiate between the
causes, and a treatment chosen on the basis of a test result) to create patient-specific knowledge bases
(PSKBs). Finally, we will explore the practicality of creating PKSBs and their usability by creating PDSs and
PKSBs for actual patients being seen by medical residents in clinic and providing the residents with a
navigational tool that makes use of the knowledge base to help them better understand their patients’ cases.
Evaluatio...

## Key facts

- **NIH application ID:** 10458471
- **Project number:** 5R21LM013401-02
- **Recipient organization:** UNIVERSITY OF ALABAMA AT BIRMINGHAM
- **Principal Investigator:** JAMES J CIMINO
- **Activity code:** R21 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2022
- **Award amount:** $161,494
- **Award type:** 5
- **Project period:** 2021-08-01 → 2023-07-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10458471

## Citation

> US National Institutes of Health, RePORTER application 10458471, Improving Electronic Health Record Usability and Usefulness with a Patient-Specific Clinical Knowledge Base (5R21LM013401-02). Retrieved via AI Analytics 2026-05-24 from https://api.ai-analytics.org/grant/nih/10458471. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*