Enhancing Healthcare Triage

The central issue in triage services lies in the need for accurate and timely assessment of patients' medical conditions. This evaluation is essential for improving patient outcomes and healthcare system efficiency. However, inaccurate triage can lead to a range of issues, including delays in care and the provision of low-value services (1). For example, up to two-thirds of emergency department visits are estimated to be unnecessary and avoidable, resulting in excessive healthcare spending at the national level. This underscores the critical need to implement more effective and precise triage systems in healthcare settings (2)(3).

‍

Size of the Problem:

‍

• Up to two-thirds of emergency department visits are unnecessary and avoidable (3).
• This results in an excess of $32 billion in healthcare spending at the national level (2).
• Emergency department wait times are increasing, with a growing number of patients waiting over 12 hours from admission to decision (1).
• Gastroenterology consultants at the Western General Hospital in Edinburgh triage approximately 30 to 40 referrals per day (4).
• The triage process is complex, with over 120 permutations of identified outcomes (4).

‍

The problem of inaccurate triage in healthcare services represents a significant concern as it leads to treatment delays, the provision of low-value care, and excessive healthcare spending nationally. This situation is further exacerbated by increasingly long wait times in emergency departments, with a growing number of patients waiting over 12 hours from admission to decision. Additionally, the complexity of the triage process in specialties like gastroenterology underscores the urgent need to implement more effective and precise systems. Improving the accuracy and efficiency of triage would not only benefit patients by enhancing their health outcomes but also alleviate the burden on healthcare professionals and reduce costs associated with healthcare.

To address the challenges in triage services, the implementation of Artificial Intelligence (AI) solutions can be a highly effective strategy. Below, we explore how AI can automate and enhance the triage process:

‍

• Automated Priority Classification: Machine learning algorithms can analyze clinical data, symptoms, and medical histories to automatically assign priority to each patient. These models can identify patterns and warning signals, helping healthcare professionals make more informed decisions, thus optimizing the triage process and improving patient care. The implementation of the Referencing and Intelligent Triage System (RITA) at NHS Lothian is an outstanding example of how artificial intelligence (AI) can automate the triage process with significant impact. RITA, using natural language processing and machine learning algorithms, has demonstrated reduced waiting times in waiting lists and administrative burden. Since its implementation in January 2020, RITA has managed to automate between 40% and 50% of urgent cancer suspicion referrals, contributing to faster and more accurate care for patients in complex medical settings (4).

• Real-Time Data Analysis: Highlights how Artificial Intelligence (AI) can process large amounts of data in real-time, constantly monitoring vital signs and alerting medical staff to significant changes. A notable example is the Electronic Triage System (ETS), recently developed and based on clinical data, designed to improve patient severity distribution in the emergency department. In a study conducted by Smith et al. in 2019, including 25,198 adult visits, ETS demonstrated more equitable distribution of patients across different severity levels and more effectively identified those with serious outcomes, with an area under the curve (AUC) of 0.83, compared to 0.73 of the current standard, the Emergency Severity Index (ESI). This example emphasizes how real-time data analysis can improve triage accuracy in emergencies, ensuring adequate resource distribution and timely care for critical patients (2).

• Resource Optimization: Artificial Intelligence (AI) can optimally allocate resources in clinical settings, minimizing bottlenecks and ensuring equitable care. For example, according to a study conducted by Yoshihiko Raita and colleagues, AI models can accurately predict the need for critical care or hospitalization in patients. These models achieved a 15% improvement in efficient resource allocation, reducing wait times and ensuring patients receive adequate care in a timely manner (3).

• Clinical Decision Support: AI systems can provide evidence-based recommendations for treatment and referral, assisting physicians in making more informed decisions and avoiding errors. MayaMD, an AI-based application, exemplifies clinical decision support by helping patients determine where to seek medical attention. In a comparative study, MayaMD matched consensus triage decisions by physicians in an impressive 92% of cases, surpassing the agreement rate among individual medical professionals. This highlights its potential to improve the quality and efficiency of emergency medical care (1).

• Chatbots and Virtual Assistants: The assistant described in this article was specifically designed to help clinical staff, including doctors and nurses, optimize triage processes in emergency departments. This system is programmed to interact efficiently and accurately with clinical staff, gathering essential information about patients' symptoms and medical conditions. It uses advanced natural language processing and machine learning algorithms to analyze the data provided, assess the urgency of each situation, and prioritize care appropriately. This assistant not only improves the accuracy of patient classification but also provides clear recommendations on the next steps, helping clinical staff guide patients to appropriate care based on the severity of the case. Its training involved compiling two comprehensive databases, which include triage protocols and assessments of medical conditions, providing a robust knowledge base for optimal functioning (6)(7).

• National Hospital and Ambulatory Medical Care Survey (NHAMCS): Provides detailed data on emergency department visits and outpatient consultations in the United States, including demographic information, medical conditions, procedures performed, and care outcomes.

‍

• Electronic Health Records (EHR): Comprehensive records containing patients' medical history, diagnoses, treatments, laboratory test results, etc., offering valuable insights into patient needs and clinical outcomes prediction.

‍

• Clinical Data Warehouses: Store clinical data from various sources such as hospital information systems, electronic medical records, and connected medical devices, providing a wide range of information for triage analysis and resource optimization.

‍

• Public Health Databases: Sources like the Centers for Disease Control and Prevention (CDC) or the World Health Organization (WHO) offer epidemiological data, disease trends, and health statistics at global and national levels, relevant for emergency resource planning and management.

‍

• Research Databases: Platforms such as PubMed, MEDLINE, or Scopus contain a wealth of clinical studies, systematic reviews, and meta-analyses, offering insights into triage practices, outcome prediction models, and resource management strategies.