Longitudinal Database System for Rapid Medical Data Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current database systems, including relational and Non-SQL databases, are inefficient in handling and analyzing large volumes of complex patient event data, which are sparse in time and have varying properties, leading to slow turnaround cycles and inability to provide rapid cycle analytics effectively.
Innovation Solution
A patient-level longitudinal database system that uses a highly-compressed longitudinal storage format and a variable scripting language to efficiently process and analyze complex medical data, enabling rapid cycle analytics by transforming and normalizing data into a time-ordered sequence, and providing a flexible query mechanism for evidence-based analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If current database systems (relational and Non-SQL) are used to store and query patient event data, then data can be stored, but query and analysis efficiency deteriorates due to the complexity and sparsity of the data over time
Solution Approach 1:
The patent segments patient event data into discrete events with specific attributes, organizing them in a longitudinal database where each patient has a timeline of events. This segmentation allows efficient querying by event type, time period, or patient cohort without processing entire datasets, directly improving data handling efficiency while reducing analysis time.
Solution Approach 2:
The patent introduces a time dimension by organizing data as a longitudinal database with events ordered chronologically for each patient. This temporal structuring enables efficient time-based queries and rapid cycle analytics by allowing direct access to events within specific time windows, rather than scanning through unstructured or horizontally organized data.
2Adaptability or versatility
If complex medical data with varying properties and nested sub levels is stored in traditional formats, then comprehensive data can be captured, but flexibility to handle and assemble data deteriorates
Solution Approach 1:
The patent implements a dynamic data model where event attributes and nested structures can adapt to different event types and data sources. The longitudinal database structure allows flexible addition of new event types, attributes, and nested elements without requiring schema changes, enabling the system to handle varying data properties while maintaining organizational efficiency.
Solution Approach 2:
The patent creates a universal data structure that can accommodate multiple types of medical events (diagnoses, procedures, medications, outcomes) with varying properties and nested sub-levels. This unified longitudinal format allows the same database structure to handle diverse data types, improving adaptability while the standardized event framework reduces overall system complexity.
3Reliability
If large volumes of sparse patient event data are stored in unstructured formats, then all data can be retained, but ability to provide reproductive or auditable analytics deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-structuring patient event data in a standardized longitudinal format during data ingestion, organizing events chronologically with consistent attributes and nested structures. This pre-processing eliminates the need for complex data assembly during analytics, ensuring reproducible results while reducing the complexity of data preparation for auditable analyses.
Data Source
AI summary
In one embodiment, a system is provided. The system includes a memory device for storing templates; and a processing device, operatively coupled to the memory device, to select a template in view of user input. The template includes one or more embedded data fields associated with a study. A query comprises a tree of operators in view of the embedded data fields is derived. The set of operators being adapted to evaluate events data stored in a longitudinal database. Results corresponding to the query are generated by applying at least one operator of the tree of operators to the events data. The results including a time series of outcomes related to the events data. Thereafter, the template is populated with a least a portion of the results for presentation in connection with the study.