Abstract
In the modern era of antibiotics, healthcare-associated infections (HAIs) have emerged as a prominent and concerning health threat worldwide. Implementing an electronic surveillance system for healthcare-associated infections offers the potential to not only alleviate the manual workload of clinical physicians in surveillance and reporting but also enhance patient safety and the overall quality of medical care. Despite the widespread adoption of healthcare-associated infections surveillance systems in numerous hospitals across China, several challenges persist. These encompass incomplete coverage of all infection types in the surveillance, lack of clarity in the alerting results provided by the system, and discrepancies in sensitivity and specificity that fall short of practical expectations. We design and develop a knowledge-based healthcare-associated infections surveillance system (KBHAIS) with the primary goal of supporting clinicians in their surveillance of HAIs. The system operates by automatically extracting infection factors from both structured and unstructured electronic health data. Each patient visit is represented as a tuple list, which is then processed by the rule engine within KBHAIS. As a result, the system generates comprehensive warning results, encompassing infection site, infection diagnoses, infection time, and infection probability. These knowledge rules utilized by the rule engine are derived from infection-related clinical guidelines and the collective expertise of domain experts. We develop and evaluate our KBHAIS on a dataset of 106,769 samples collected from 84,839 patients at Gansu Provincial Hospital in China. The experimental results reveal that the system achieves a sensitivity rate surpassing 0.83, offering compelling evidence of its effectiveness and reliability.
Introduction
Healthcare-associated infections (HAIs) represent a significant global health threat, increasing patient mortality and adding billions in healthcare costs. In China, many hospitals still rely on manual reviews of patient records—a process that is labor-intensive and prone to delays. While automated systems exist, they often struggle with outdated diagnostic criteria, data silos (especially unstructured text in progress notes), and a "sensitivity gap" that leads to missed infections or untrustworthy alerts. This paper introduces KBHAIS, a knowledge-based surveillance system designed to address these challenges by integrating structured clinical data with expert heuristics.
Methods
System Architecture
The KBHAIS system utilizes a modular architecture centered on a Rule-Based Reasoning (RBS) engine. It extracts data from four primary hospital systems: HIS (Hospital Information System), LIS (Laboratory Information System), PACS (Picture Archiving and Communication System), and EMR (Electronic Medical Record).
The core of the system is the Reasoning Engine, which employs the Drools business rule engine. This engine processes data through a hybrid knowledge base consisting of 154 formal clinical guideline rules and 52 expert-derived "common understanding" rules. This combination allows the system to capture infections that might be missed by formal guidelines alone, such as those indicated by specific combinations of symptoms and keywords in progress notes.
Case Representation
To facilitate automated reasoning, patient visits are transformed into a standardized 8-tuple format. This representation captures a wide array of clinical factors, ranging from demographic data and vital signs to complex imaging reports and medication orders.
Results and Key Insights
The system was validated using a massive real-world dataset of 106,769 samples from 84,839 patients at Gansu Provincial Hospital. KBHAIS achieved an average sensitivity of 0.83, with particularly high performance in Bloodstream Infections (BSI) and Gastrointestinal infections (AD/GI), both exceeding 90%.
Key insights from the development include:
- The Expert Gap: Clinical guidelines for certain infections often lag behind actual clinical practice. Integrating 52 expert rules significantly boosted the system's ability to catch cases missed by formal guidelines.
- Managing Alert Fatigue: To prevent "double-counting," the system uses a 14-day Repeat Infection Timeframe (RTI). If a patient is flagged for the same infection site within 14 days, duplicate alerts are suppressed.
- Data Reliability: Sensitivity dropped during February 2021 (Chinese New Year), likely due to delays in doctors updating electronic records during busy holiday periods, highlighting the system's dependence on timely data entry.
Discussion and Future Work
While KBHAIS demonstrates high sensitivity and reliability, challenges remain in extracting subjective symptoms from unstructured text using keyword matching. Future iterations aim to integrate deep learning-based Natural Language Processing (NLP) to better understand the context of physician notes. Additionally, while overall sensitivity is high, precision for intraabdominal and soft tissue infections requires improvement to reduce false alarms. Future work will focus on multi-center validation to ensure the system's stability and portability across different hospital IT infrastructures.
Summary updated