Disease Characterization Using Angle-Based Metabolomic Profiles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Interpreting complex biomedical data for diagnosing steroidogenic disorders is difficult due to heterogeneous data and requires significant expertise, and existing methods struggle with missing values and rare disease conditions.
Innovation Solution
An interpretable prototype-based machine learning method using angle-based dissimilarity between metabolomic profiles, accounting for varying dimensionality and individual proportional characteristics, to identify disease states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex biomedical data is collected to improve diagnostic accuracy, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent introduces an intermediary computational system that processes complex biomedical data through multiple stages: data normalization, feature extraction, and pattern recognition algorithms. This intermediary layer transforms raw complex data into interpretable diagnostic insights, resolving the contradiction by mediating between data complexity and diagnostic accuracy.
Solution Approach 2:
The patent segments the complex biomedical data into distinct analytical components including metabolic profiles, genetic markers, and clinical parameters. Each segment is processed independently through specialized algorithms before being integrated into a comprehensive diagnosis, thereby managing complexity while maintaining high measurement precision.
2Measurement precision
If expert knowledge is required to interpret data, then measurement precision is improved, but ease of operation deteriorates
Solution Approach 1:
The patent implements self-service through automated interpretation systems that perform data analysis without requiring expert intervention. The system automatically normalizes data, identifies patterns, and generates diagnostic recommendations, enabling non-experts to access high-precision interpretation capabilities through an user-friendly interface.
Solution Approach 2:
The patent replaces the mechanical system of expert human analysis with computational algorithms and machine learning models. These automated systems perform the interpretive function previously requiring expert knowledge, thereby maintaining measurement precision while dramatically improving ease of operation and accessibility.
3Reliability
If traditional methods are used to handle missing values, then reliability deteriorates, but device complexity increases
Solution Approach 1:
The patent applies parameter changes by transforming missing value handling from traditional imputation methods to a probabilistic framework. The system models missing data as a specific parameter state and uses probability distributions to account for uncertainty, thereby improving diagnostic reliability without requiring complex ad-hoc processing for each missing value scenario.
Data Source
Figure 1
Figure 1
Figure 1
AI summary
The invention provides a method of characterising a disease state comprising: (i) collecting metabolic data from a plurality of subjects; (ii) presenting the data as vectors with dimensions corresponding to different biomarkers: and (iii) weighting the importance of either individual dimensions, or the interplay among multiple dimensions when calculating angles of the vectors, such that there is a minimum variation of angle within a disease class and/or a maximum variation of angle compared to a different disease class. The invention also describes a method of identifying a disease state or following progression of a disease state in a subject comprising: (i) collecting metabolic data from the subject; (ii) presenting the data as vectors with dimensions corresponding to different biomarkers: and (iii) comparing two or more angles of vectors with a prototype vector and optionally at least one relevance matrix, to identify the presence of, or progression of, a disease state.