Information Extraction From Electronic Health Records to Predict Readmission Following Acute Myocardial Infarction: Does Natural Language Processing Using Clinical Notes Improve Prediction of Readmission?

Abstract

Background Social risk factors influence rehospitalization rates yet are challenging to incorporate into prediction models. Integration of social risk factors using natural language processing (NLP) and machine learning could improve risk prediction of 30‐day readmission following an acute myocardial infarction. Methods and Results Patients were enrolled into derivation and validation cohorts. The derivation cohort included inpatient discharges from Vanderbilt University Medical Center between January 1, 2007, and December 31, 2016, with a primary diagnosis of acute myocardial infarction, who were discharged alive, and not transferred from another facility. The validation cohort included patients from Dartmouth‐Hitchcock Health Center between April 2, 2011, and December 31, 2016, meeting the same eligibility criteria described above. Data from both sites were linked to Centers for Medicare & Medicaid Services administrative data to supplement 30‐day hospital readmissions. Clinical notes from each cohort were extracted, and an NLP model was deployed, counting mentions of 7 social risk factors. Five machine learning models were run using clinical and NLP‐derived variables. Model discrimination and calibration were assessed, and receiver operating characteristic comparison analyses were performed. The 30‐day rehospitalization rates among the derivation (n=6165) and validation (n=4024) cohorts were 15.1% (n=934) and 10.2% (n=412), respectively. The derivation models demonstrated no statistical improvement in model performance with the addition of the selected NLP‐derived social risk factors. Conclusions Social risk factors extracted using NLP did not significantly improve 30‐day readmission prediction among hospitalized patients with acute myocardial infarction. Alternative methods are needed to capture social risk factors.

Artificial intelligence: revolutionizing cardiology with large language models

“…The identification and characterization of cardiovascular disease cohorts, signs and symptoms of disease, reduction of missingness, and assessment of risk factors and comorbidities are a few examples regarding the phenotypic assessment of patients. 5 , 17–23 Ultimately, supplementing structured information (lab, medications, vitals, and codes) with information derived from unstructured data is likely to improve patient phenotyping. Through real-time phenotyping, relevant information on patient’s clinical status can be provided in dashboards to be used for clinical decision support or to aid phenotype harmonization like the HDR-UK phenotype library.…”

“…The latest advance in language generation interfaces took the world by storm in just a few weeks, creating full-length documents, poems, and code almost indistinguishable from human-generated content generated from short prompts and questions. These interfaces, such as OpenAI’s ChatGPT, 5 , 6 based on the GPT family of language models (LMs), and Google’s Bard, 7–9 based on the PaLM2 model, have led to mistaken claims 10–12 that ChatGPT has passed what in 1950 was defined as the ultimate test of artificial intelligence (AI)—the Turing test 13 —whereby a computer programme could fool a human into thinking that a dialogue interaction with it was actually with another human. Despite these claims, even though ChatGPT can imitate interaction that is almost indistinguishable from interaction with a human, true dialogue interaction has not yet been achieved as that would require understanding of physical and psychological laws, thought processes and connections of ideas, logics, beliefs, and values that are beyond what ChatGPT is currently able to achieve ( Figure 1A ).…”

“… 14–16 Specifically in cardiology, NLP has been proposed and tested for the identification and characterization of cardiovascular disease cohorts, recognition of signs, symptoms, risk factors, comorbidities, and medical reasoning. 5 , 17–23 Additionally, from free text reports, measurements not recorded in a structured manner can be obtained.…”