An expert rule-based approach for identifying infantile-onset Pompe disease patients using retrospective electronic...

Area of Science:

• Medical Genetics
• Rare Diseases
• Public Health Informatics

Background:

• Pompe disease is a rare genetic disorder causing glycogen accumulation, severely impacting the heart and muscles.
• Infantile-onset Pompe disease (IOPD) demands prompt treatment to prevent mortality, yet diagnostic delays are common due to limited resources.
• Early and accurate diagnosis is critical for effective management and improved outcomes in IOPD patients.

Purpose of the Study:

• To develop and evaluate an expert rule-based screening approach using electronic health records (EHRs) for early detection of IOPD in the UAE.
• To streamline the diagnostic process for IOPD, enabling faster identification and initiation of treatment.
• To leverage existing healthcare data infrastructure for improved rare disease diagnosis.

Main Methods:

• Utilized EHR data from the Abu Dhabi Healthcare Company (SEHA) network in the UAE.
• Developed an expert rule-based screening system integrated into a dashboard for automated patient identification.
• Defined expert rules based on age, specific symptoms, and creatine kinase levels to identify high-risk IOPD cases.
• Evaluated the screening approach using accuracy, sensitivity, and specificity metrics.

Main Results:

• The screening approach identified five true positive IOPD cases, one false negative, and four false positives from 93,365 subjects.
• The false negative case highlighted challenges in diagnosing co-occurring conditions and the importance of creatine kinase measurements.
• False positive cases were attributed to other genetic disorders and infections, indicating the need for differential diagnosis.
• The rule-based dashboard facilitated efficient data visualization and automated screening for potential IOPD patients.

Conclusions:

• Integrating expert rules with EHRs and a dashboard offers an efficient method for automated patient screening and early detection of IOPD.
• The developed approach supports timely intervention, potentially improving patient outcomes for this rare genetic disorder.
• Future research should explore machine learning to further enhance the precision and efficiency of IOPD identification.