Utilising NLP integrated with microservices architecture to analyse longitudinal medical records to reconstruct comprehensive patient journeys

About the Project

Electronic Medical Records (EMRs) provide a streamlined and secure digital platform, facilitating the storage, management, and sharing of patient information among healthcare providers. The adoption of EMRs not only enhances care coordination and reduces errors but also empowers healthcare professionals to make more informed decisions, ultimately leading to improved overall healthcare outcomes. During consultations, data is recorded in EMRs using a structured format, often complemented by clinical coding systems and as free-text entries. The structured data elements play a pivotal role in ensuring consistency in recording medical information, making it easier to retrieve, analyse, and exchange among providers and systems. Various incentivised and quality-driven approaches have been developed to improve the utility of structured data, particularly to support chronic disease management, monitoring, and epidemiology research. Despite the focus on structured data, the complexity of computer-mediated clinical consultations demands the inclusion of free-text medical narratives in EMRs. Free-text entries offer a more human-readable narrative, capturing the nuances and context of a patient's health information. They are essential for documenting information that may not neatly fit into clinical coding systems, allowing clinicians to record unique aspects of a patient's medical history, personal circumstances, and lifestyle. This contributes to a more personalised and holistic understanding of individual healthcare needs, providing flexibility and adaptability in the evolving landscape of medical knowledge and the diversity of patient stories.

In clinical consultations, clinicians often refer to past free-text entries to obtain contextual information in addition to coded data on problems and symptoms. The combined longitudinal free-text narratives can reveal significant medical or lifestyle-related information, offering a more holistic overview. Moreover, free-text entries has the potential role in validating structured data, enhancing its quality, and aiding in error detection. The temporal aspects captured in free text, including presenting problems, patient expectations, concerns, and symptom progressions, provide valuable insights into a patient's health journey.

Significant achievements have been made in extracting clinical information from free text, including symptoms, diseases, and procedures, through Natural Language Processing (NLP). NLP has also been used to support clinical decision support systems and facilitate semantic interoperability by converting free text into structured data.

This project aims to leverage NLP to review longitudinal medical records, adopting a patient-oriented and epidemiological evidence supported approach The proposed project involves integrating NLP with microservices architecture to enhance the flexibility, scalability, and modularity of NLP applications in EMR systems. The underlying belief is that adopting a microservices-based modular approach would improve the accuracy, context awareness, and interoperability of NLP tasks, given the intricate nature of medical knowledge, associated ontologies, and the diverse patient journeys. The overarching objective of the project is to advance patient-oriented and evidence-based care by developing clinical systems that are less burdensome and more context-aware, thereby promoting the psychosocial aspects of computer-mediated consultations.