Identifying, Analyzing, and Visualizing Diagnostic Paths for Patients with Nonspecific Abdominal Pain

Diagnosing abdominal pain remains a clinical challenge due to its nonspecific nature. While prior research has shown that symptoms like abdominal pain often lead to diagnostic uncertainty, no prior work had resolved how to systematically track and analyze diagnostic steps in real-world settings. Existing studies have focused on individual diagnostic tools or clinical guidelines, but none have used electronic health records to map full diagnostic pathways. This gap motivated the use of process-mining techniques to extract and visualize diagnostic paths from EHR data. Prior knowledge indicated that abdominal pain often leads to diagnostic delays, but no prior work had resolved how to quantify or visualize these delays. The lack of structured diagnostic pathway analysis in clinical settings created a need for new methods. This paper introduces a novel approach using EHR data to construct and analyze diagnostic paths. The absence of such a method previously limited understanding of diagnostic processes in abdominal pain. By addressing this gap, the study contributes to improving diagnostic accuracy and efficiency.

This study aimed to develop a method for identifying and visualizing diagnostic paths for patients with nonspecific abdominal pain using electronic health records. The specific problem addressed was the lack of structured diagnostic pathway analysis in clinical settings. The motivation came from the need to better understand diagnostic delays and patterns in abdominal pain cases. The study sought to extract diagnostic steps from EHR data to reveal patterns in how diagnoses are reached. The goal was to use process-mining methods to construct and analyze these diagnostic paths. The study focused on patients presenting with abdominal pain who had not received a stable diagnosis. The researchers proposed that analyzing these paths could improve diagnostic practices. The ultimate aim was to enhance both the analysis and comprehension of diagnostic processes.

The researchers used process-mining methods to extract diagnostic paths from electronic health records. Patient data included features, diagnostic actions, and recorded diagnoses for 501 adult patients with abdominal pain. Data were collected from a hospital system in suburban Chicago. A stable diagnosis was defined as a diagnosis recorded twice within 12 months. A working diagnosis was defined as a diagnosis recorded only once. Three types of path visualizations were generated from the data. The study focused on diagnostic steps taken from initial presentation until a diagnosis was obtained or the evaluation ended. The researchers used structured EHR data to construct diagnostic paths for analysis. This approach allowed for the visualization of diagnostic patterns over time.

A stable diagnosis was obtained in 63 (13%) patients after 12 months. In 271 (54%) patients, a working diagnosis was recorded. The mean diagnostic path duration was 145.3 days with a standard deviation of 195.1 days. These 63 patients received 75 stable diagnoses. Three distinct types of diagnostic path visualizations were generated from the data. The study found that structured EHR data can be used to construct diagnostic paths. The findings suggest that diagnostic paths vary in duration and complexity. The use of process-mining methods revealed patterns in diagnostic steps. These results indicate that diagnostic path analysis can provide insights into diagnostic practices. The study demonstrated that EHR data can support the visualization of diagnostic processes.

The authors propose that diagnostic path analysis can reveal patterns in how diagnoses are reached for nonspecific abdominal pain. They suggest that structured EHR data can be used to construct and visualize diagnostic paths. The findings indicate that diagnostic processes vary in duration and complexity. The study supports the use of process-mining methods in analyzing diagnostic steps. The researchers propose that such analysis can improve understanding of diagnostic practices. The results suggest that diagnostic path visualizations can enhance comprehension of the diagnostic process. The authors suggest that these methods can be applied to other diagnostic challenges. They propose that visual analytics can support both analysis and comprehension of diagnostic paths.