Background

Low back pain (LBP) is one of the most common causes of disability in US adults that impacts the ability to work and affects the quality of life. LBP events also lead to the significant financial burden for both individuals and clinical facilities, with combined direct and indirect costs of treatment for musculoskeletal injuries and associated pain estimated to be approximately $213 billion annually.

LBP events are essentially categorized into two types. Acute LBP occurs suddenly, usually associated with trauma or injury with subsequent pain, whereas chronic LBP is often reported by patients in regular checkups and has led to a significant increase in the use of healthcare services over the past two decades. It is very important to differentiate between acute and chronic LBP in the clinical setting as these conditions as well as their management and medical billing are substantially different.

However, these events are usually not explicitly separated in the electronic health records (EHRs) due to the lack of direct coding conventions. For example, the code M54.5 to characterize “Low back pain” diagnosis, does not differentiate among the various facets characterizing the condition. The acuity is usually reported in the clinical notes, requiring retrospective chart review of the free-text to characterize LBP events, which is time-consuming and not scalable.

Technology Overview

Researchers at Mount Sinai have developed a natural language based algorithm, ConvNet, to analyze the text data of the clinical notes to characterize LBP events. The algorithm was trained using over 17,000 clinical notes with an AOC of 0.98 and F-score of 0.70.

Stage of Development

Software is available.

Benefits

Automated, accurate and scalable

Applications

Improve accuracy of diagnosing lower back pain to improve treatment
Improved quality of life
Minimize financial impact on patient, companies and healthcare costs
Can be readily deployed in clinical quality improvement processes for LBP

ConvNet: A Natural Language Processing Based Machine Learning Algorithm to Characterize Lower Back Pain from Analysis of Clinical Notes