A

Clinical Decision Support

Clinical Decision Support (CDS) Notes

Introduction to Clinical Decision Support (CDS)

  • Definition: Clinical decision support (CDS) provides clinicians, staff, patients, or other individuals with knowledge and person-specific information, intelligently filtered or presented at appropriate times, to enhance health and health care (AMIA Roadmap).

  • Relationship with Computerized Physician/Provider Order Entry (CPOE):

    • CDS may be used with CPOE.

    • CDS can be used independently of CPOE.

    • CPOE may not always incorporate CDS.

Applications of CDS

CDS is applied in various tasks across healthcare:

  • During data-entry tasks:

    • Smart documentation forms.

    • Order sets, care plans, and protocols.

    • Parameter guidance.

    • Critiques and immediate warnings (alerts).

  • During data-review tasks:

    • Relevant data summaries for a single patient.

    • Multi-patient monitors.

    • Predictive and retrospective analytics.

  • During assessment and understanding tasks:

    • Filtered reference information and knowledge resources.

    • Expert workup and management advisors.

  • Not triggered by user task:

    • Event-driven alerts (triggered by data changes).

    • Time-triggered reminders.

Historical Perspectives of CDS

  • Early Focus (1970s-1980s): Primarily on the application of artificial intelligence (AI) and expert systems to improve medical diagnosis.

    • Diagnostic decision support was a major area.

    • Computer-aided diagnosis proved challenging, leading to a shift towards more focused capacities to reduce errors and improve quality.

    • These early efforts established the intellectual foundation for modern CDS and its focused approach.

  • Relevance to Modern EHRs: With the availability of data in modern Electronic Health Records (EHRs), some initial approaches might regain utility in the future.

Historical Legacy of Decision Support in Informatics

  • Artificial Intelligence (AI):

    • An area of computer science (CS) focused on building programs that exhibit characteristics of human intelligence.

    • Initial systems did not fulfill their promises.

    • Recent resurgence due to advancements in machine learning, increased data availability, and enhanced computer processing power.

  • Expert System (ES): A computer program designed to mimic human expertise.

  • Decision Support System (DSS):

    • Also mimics human expertise but plays a more supportive rather than independent role.

    • Historically, in medicine, DSS focused on:

      • Diagnostic decision support: Assisting in patient diagnosis.

      • Therapeutic decision support: Aiding in patient treatment.

Early Attempts to "Quantify" Medical Diagnosis

  • Ledley and Lusted (1959, 1960):

    • Proposed a mathematical model for diagnosis.

    • Utilized set theory and symbolic logic for clinical findings, making diagnoses based on probabilities.

  • Warner (1961):

    • Developed a mathematical model for diagnosing congenital heart disease.

    • Used a contingency table approach with diagnoses as rows and symptoms as columns.

    • The system predicted the diagnosis with the highest conditional probability for a given set of symptoms.

Approaches to Diagnostic Expert Systems (ESs)

  • The functions of these systems are closely tied to their methods of knowledge representation.

  • Four General Approaches:

    • Clinical algorithms.

    • Bayesian statistics.

    • Production rules.

    • Scoring and heuristics.

  • Current Approaches: Leverage modern EHRs and other technological advancements.

Clinical Algorithms

  • Mechanism: Follows a predefined path through a