# Evaluating and Improving Utilization of Evidence-Based Medical Therapy in Patients with Heart Failure using Automated Tools in the Electronic Health Record

> **NIH NIH K23** · YALE UNIVERSITY · 2022 · $192,046

## Abstract

PROJECT SUMMARY
Heart failure (HF) affects over 6 million US adults, with high rates of hospitalization and nearly 50% mortality at
5 years from diagnosis. Nearly half of these patients have systolic HF with multiple evidence-based therapeutic
options proven to reduce the risk of hospitalization and mortality in this subgroup of patients. Evaluating the
appropriate utilization of these therapies is currently limited to post-hoc assessments of manually abstracted
patient records at a limited number of hospitals participating in quality improvement registries. These manual
abstraction strategies do not offer opportunities to improve care in real-time, and even at hospitals engaged in
quality improvement efforts, only 1 in 5 of eligible patients with HF receive all first-line evidence based medical
treatments. In this patient-oriented mentored career development award proposal, Dr. Rohan Khera proposes to
leverage the ubiquitous digitization of medical records in the electronic health record (EHR) to address the
adequate utilization of evidence based medical therapy in HF. He proposes to use a large, publicly accessible,
deidentified EHR database to develop and validate an algorithm that uses deep learning based natural language
processing (NLP) within unstructured clinical documentation for hospitalized HF patients to identify those with
systolic HF (Aim #1). He will engage clinicians to design consensus-based algorithms to identify
contraindications to HF treatments, developed as algorithms within the EHR (Aim #2). Finally, he will construct
a prototypic clinical decision support (CDS) tool identifying HF treatment eligibility in real-time using the
algorithms and evaluate potential implementation strategies using qualitative evaluation of feedback from
clinicians and patients (Aim #3). While proposed as a strategy to evaluate quality of care of individual patients,
the proposed research will also model a fully automated electronic clinical quality measure for HF. The algorithms
will be made open source to allow institutions to validate and apply them to their individual care setting. The
proposal is supported by strong mentorship from experts in quality measure design, informatics, advanced NLP,
CDS design, and qualitative research methodology. The facilities at Yale Center of Outcomes Research and
Evaluation, which designs and evaluates national quality measures, and has access to computational resources
required to accomplish the research goals as well as to the Yale EHR to validate the models are major strengths
of the application. The proposed period of mentored research will support Dr. Khera’s training in medical
informatics, advanced analytic tools such as NLP, and qualitative research methodology. The experience and
skillset acquired during this period will support Dr. Khera’s transition to independence where he plans to lead
multi-institutional collaboratives to evaluate the use of automated tools in the measurement and improvement of
th...

## Key facts

- **NIH application ID:** 10375578
- **Project number:** 5K23HL153775-02
- **Recipient organization:** YALE UNIVERSITY
- **Principal Investigator:** Rohan Khera
- **Activity code:** K23 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2022
- **Award amount:** $192,046
- **Award type:** 5
- **Project period:** 2021-04-01 → 2026-03-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10375578

## Citation

> US National Institutes of Health, RePORTER application 10375578, Evaluating and Improving Utilization of Evidence-Based Medical Therapy in Patients with Heart Failure using Automated Tools in the Electronic Health Record (5K23HL153775-02). Retrieved via AI Analytics 2026-05-24 from https://api.ai-analytics.org/grant/nih/10375578. Licensed CC0.

---

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