Related Experiment Video
Updated: Jul 4, 2026
10:26
A Multicenter MRI Protocol for the Evaluation and Quantification of Deep Vein Thrombosis
Published on: June 2, 2015
Multisite Real-World Validation of an Electronic Health Record-Integrated Generative Artificial Intelligence Tool for
Medrxiv : the Preprint Server for Health Sciences
|July 3, 2026
Summary
A new generative AI tool (GenAI) shows higher sensitivity for identifying venous thromboembolism (VTE) risk in hospitalized patients compared to current methods. This AI can aid clinicians in risk stratification, improving patient care.
Area of Science:
- Medical Informatics
- Artificial Intelligence in Healthcare
- Clinical Decision Support
Background:
- Venous thromboembolism (VTE) risk stratification is crucial for appropriate inpatient thromboprophylaxis, but current methods lack consistency in routine clinical practice.
- Health systems are exploring artificial intelligence (AI) tools, yet rigorous evaluations within learning health systems (LHS) are limited.
- This study evaluates an electronic health record (EHR)-integrated generative AI (GenAI) system, inHealth General Reasoner (iHGR), for VTE risk stratification.
Purpose of the Study:
- To compare the VTE risk stratification performance of a pilot GenAI system (iHGR) against clinician order set classifications and physician-adjudicated chart review.
- To assess the sensitivity and specificity of the iHGR system for VTE risk stratification.
- To analyze workflow comparators and error patterns between the AI and clinician-based methods.
Main Methods:
- A multisite retrospective validation study involving 500 adult inpatient admissions at Johns Hopkins Medicine.
- Comparison of iHGR classifications against a reference standard (physician-adjudicated chart review).
- Evaluation of clinician-selected order sets (checklist-based and clinician judgment-based) against the same reference standard.
Main Results:
- The iHGR system achieved 81.8% sensitivity and 70.9% specificity for VTE risk stratification.
- Checklist-based order sets showed lower sensitivity (61.3%) but higher specificity (86.2%).
- Clinician judgment-based order sets had 78.1% sensitivity and 65.4% specificity; false negatives in iHGR were linked to missed narrative risk factors.
Conclusions:
- The iHGR system demonstrated superior sensitivity for VTE risk compared to existing order sets without introducing systematic bias.
- In silico evaluation of AI within LHSs can identify performance trade-offs before large-scale implementation.
- GenAI shows promise in supporting clinician judgment for VTE risk stratification, rather than replacing it, with narrative data abstraction remaining a key area for improvement.
Related Concept Videos
01:29Venous Thrombosis III: Interprofessional Care
Venous thrombosis requires effective prevention and treatment strategies to improve patient outcomes and reduce potential complications.Prevention StrategiesHealthcare providers must prioritize preventing venous thromboembolism (VTE) for all adult patients upon admission. Interventions depend on bleeding and thrombosis risk, medical history, current medications, diagnoses, planned procedures, and patient preferences. Patients on bed rest should change positions every two hours and, if not...
01:20Venous Thrombosis II: Clinical Manifestations and Diagnostic Studies
The key difference between Superficial Vein Thrombosis (SVT) and Deep Vein Thrombosis (DVT) lies in their location and severity.Clinical ManifestationsSVT typically presents with localized pain, tenderness, and redness along the course of a superficial vein, often accompanied by a palpable, cord-like structure under the skin. This condition is usually less dangerous than DVT but can be uncomfortable and may lead to complications such as cellulitis or, rarely, a clot extension into the deep...
01:30Venous Thrombosis I: Introduction
Venous thrombosis, the most common disorder of the veins, involves the formation of a thrombus or blood clot associated with vein inflammation. It can be classified as either superficial vein thrombosis or deep vein thrombosis.Superficial Vein Thrombosis: This involves the formation of a thrombus in a superficial vein, usually the greater or lesser saphenous vein. Though less severe than deep vein thrombosis (DVT), SVT can lead to complications if untreated.Deep Vein Thrombosis (DVT): This...
01:27Nursing Clinical Information System
Nursing Clinical Information System (NCIS)
A Nursing Clinical Information System (NCIS) is a specialized type of healthcare information system tailored to meet the unique needs of nursing practice. It incorporates the principles of nursing informatics to streamline information management and improve the quality of care delivery.
Critical attributes of NCIS include:
A Nursing Clinical Information System (NCIS) is a specialized type of healthcare information system tailored to meet the unique needs of nursing practice. It incorporates the principles of nursing informatics to streamline information management and improve the quality of care delivery.
Critical attributes of NCIS include:
01:30Health Information Technology and Healthcare Information System
Health Information Technology (HIT)
Health Information Technology, commonly called HIT, integrates advanced information systems and technology in healthcare settings. Its primary functions include:
Health Information Technology, commonly called HIT, integrates advanced information systems and technology in healthcare settings. Its primary functions include:
01:30Venous Thrombosis IV: Nursing Management
Nursing management begins with a thorough assessment of the patient's health history. Key factors include trauma to veins, peripherally inserted central catheters, varicose veins, recent pregnancy or childbirth, surgery, bacteremia, prolonged bed rest, atrial fibrillation, COPD, heart failure, cancer, coagulation disorders, myocardial infarction, spinal cord injury, stroke, prolonged travel, recent bone fractures, and dehydration. Review medication intake, particularly oral contraceptives,...