Nature Sub-Issue: New Automated Early Idiopathic Pulmonary Fibrosis Screening Tool

According to researchers from Will Cornell Medical School, New York Presbyterian Hospital, the University of Chicago, Brigham Women's Hospital, and the Mayo Clinic, a new electronic medical record screening tool can accurately identify high-risk patients
with or develop progressive lung scarring, a condition known as idiopathic pulmonary fibrosis.

Idiopathic pulmonary fibrosis (IPF) is often fatal, in part because it is often diagnosed late in the course of the disease, when existing treatments are less effective
.
A paper published Sept.
29 in the journal Nature Medicine describes this new screening tool, which has the potential for early detection of IPF routines
.

Study co-author Dr Fernando Martinez said: "Having a robust screening approach is a major advance
in ensuring early diagnosis based on easily available parameters in electronic medical records.
"

Dr.
Ishanu Chattopadhyay, a co-author of the study and an assistant professor of medicine at the University of Chicago, said: "This tool requires little additional cost [of the patient's time] to identify the characteristics
of the disease before symptoms appear.
As long as someone arranges to see their primary care doctor, the program can run screening tools to get results
before the patient walks into the clinic.
”

The IPF affects tens of thousands of people in the United States, as well as millions of people around the
world.
It is characterized by a gradual decline in lung function, which converts healthy lung tissue into scarred fibrous tissue
due to chronic pneumonia.
The triggers for IPF are "idiopathic" — unknown — but risk factors include age, male sex, and smoking history
.
IPF has anti-fibrotic drug treatments that can prolong the life of
patients.
However, these treatments tend to be more effective early in the course of the disease, while IPF is usually diagnosed several years after symptoms appear
.
These symptoms are variable and nonspecific, but usually include shortness of breath that progresses slowly due to exertion, which patients may not notice or may be ignored
as part of natural aging.
In addition, the tests required to confirm IPF are cumbersome and expensive, which often makes patients reluctant to undergo tests
until symptoms become apparent.

"Traditionally, the diagnosis of IPF has required a multidisciplinary approach, including pulmonary clinicians, radiologists, laboratory and pathology specialists," said
Dr.
Andrew Limper, a professor of medicine and co-author of the study and director of the Mayo Clinic's Department of Lung and Critical Care Medicine.

The new software screening tool is designed to automatically detect high IPF risk based on a patient's electronic health record, simplifying and speeding up the diagnostic process
.
In principle, it will be used primarily by primary care physicians who will refer patients identified by algorithms to tests that require confirmation or exclusion of IPF
.

Dr Gary "Matt" Hunninghake, Director of the Interstitial Lung Disease Program at Brigham Women's Hospital and co-author, said: "This work is novel because the information already captured in the medical records is being used to identify patients in the system who
may be at higher risk.
"

The software's core algorithm was developed using a machine learning process to calculate the IPF risk score (IPF zero-burden co-morbidity risk score, or ZCoR-IPF) by looking at known IPF-related factors and IPF-related events (or even sequences of IPF-related events)
in a patient's health record over the past two years.

The team trained the algorithm on a commercial insurance claims database covering millions of patients in the U.
S.
from 2003-18 and then validated
it using three additional claims datasets.
In all, the development of the tool is based on the records of nearly 3 million patients, including more than 54,000 metrical cases
.

Validation trials have shown that the tool is quite sensitive and specific in identifying patients at high risk of IPF
.
Those with a risk score above the detection threshold were more than 30 times more
likely to receive an IPF diagnosis in the following year compared to those who were not screened.

The researchers now hope to expand the tool to primary care centers, where it can be evaluated and formally trialed
in real-world settings.
They also hope that the same approach, which combines machine learning algorithms and electronic health records, could be applied to other diseases, and that earlier diagnoses could save many lives
.

This approach is a paradigm shift
in IPF screening.