EMR2vec: Bridging the gap between patient data and clinical trial.

Area of Science:

• Biomedical Informatics
• Computational Biology
• Public Health

Background:

• Life-threatening viral diseases (e.g., SARS, Ebola, COVID-19) necessitate rapid clinical research and effective outbreak response.
• Data integration of patient electronic health records (EHR) and clinical trial data offers a powerful approach to accelerate research and improve patient care.
• Existing methods for matching patients to clinical trials can be inefficient, hindering timely access to experimental treatments.

Purpose of the Study:

• To introduce EMR2vec, an innovative platform designed to link patients with suitable clinical trials.
• To leverage advanced Natural Language Processing (NLP), machine learning, and semantic web techniques for enhanced data integration.
• To accelerate clinical research and improve personalized patient care, particularly in the context of emerging infectious disease outbreaks.

Main Methods:

• Developed EMR2vec platform utilizing NLP, machine learning, and semantic web technologies.
• Derived a 'bag of medical terms' (BoMT) from clinical trial eligibility criteria, normalizing entities via SNOMED-CT ontology.
• Employed ontological reasoning to represent EHR and clinical trial data in a vector space model.
• Implemented a matching process involving neural network-based dimensionality reduction and orthogonality projection for vector similarity measurement.
• Evaluated the EMR2vec platform using a Big data-based prototype.

Main Results:

• The EMR2vec platform successfully links potential patients to relevant clinical trials.
• The BoMT and ontological reasoning approach effectively represents complex medical data.
• The vector space model and similarity measurement facilitate efficient patient-trial matching.
• The Big data prototype demonstrates the scalability and effectiveness of the EMR2vec system.

Conclusions:

• EMR2vec enhances the ability of clinical researchers to respond to infectious disease outbreaks by integrating patient and trial data.
• The platform enables clinicians to identify suitable clinical research opportunities for patients or to select patients for personalized care.
• EMR2vec represents a significant advancement in utilizing data integration and AI for accelerating clinical research and improving public health outcomes.