Gene Expression Cancer Classification Using Recursive Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current cancer diagnosis methods, particularly for childhood cancers, are often delayed, incorrect, or missed due to differences between adult and childhood tumors, and lack comprehensive molecular assays for accurate classification.
Innovation Solution
A computer-implemented method and system using recursive clustering and machine learning models, such as convolutional neural networks and Gradient Boosting ensembles, to classify cancer types from gene expression data, incorporating features like variance removal, non-linear dimensionality reduction, and hierarchical clustering to identify tumor subtypes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If genome sequencing is used to detect cancer fusions, then diagnostic markers can be identified, but the method has limited diagnostic utility beyond these markers and cannot comprehensively diagnose all pediatric cancers
Solution Approach 1:
The patent develops a universal machine learning classifier that can diagnose multiple types of pediatric cancers simultaneously using gene expression data. The system is trained on diverse cancer types including leukemia, lymphoma, neuroblastoma, and sarcoma, enabling it to comprehensively classify tumors across different origins and characteristics, thereby achieving multi-functionality in cancer diagnosis
2Ease of manufacture
If adult cancer classification methods are applied to childhood cancers, then existing diagnostic frameworks can be used, but accuracy decreases due to fundamental differences between adult and childhood tumor biology
Solution Approach 1:
The patent creates specialized classification models tailored to childhood cancer characteristics. The system incorporates features specific to pediatric tumors such as embryonal origin markers, childhood-specific fusion genes, and age-related expression patterns. This local customization ensures the diagnostic tool accurately captures the unique biology of childhood cancers rather than forcing them into adult cancer classification frameworks
3Measurement precision
If comprehensive molecular assays are developed for all pediatric cancers, then diagnostic accuracy improves, but the complexity and cost of the system increases
Solution Approach 1:
The patent transforms the cancer diagnosis approach by changing from detecting specific molecular markers to analyzing overall gene expression patterns. The machine learning classifier processes high-dimensional gene expression data to identify diagnostic patterns, thereby achieving comprehensive cancer classification through a unified computational framework rather than multiple separate molecular assays
4Ease of operation
If traditional cancer diagnosis methods are used, then existing protocols can be followed, but diagnosis is frequently delayed, incorrect, or missed entirely
Solution Approach 1:
The patent replaces manual pathological examination and traditional diagnostic workflows with an automated machine learning classification system. The computational model rapidly analyzes gene expression data and provides diagnostic classifications, eliminating the time delays and subjectivity inherent in traditional methods while maintaining ease of use through standardized input requirements
Data Source
AI summary
The present invention relates to systems and methods for classification of cancer from gene expression input data. The method including: receiving a training dataset including nucleic acid data points from one or more samples; identifying classes in the training dataset including performing recursive clustering by, at each successive iteration, performing a search to identify clusters based on similarity, wherein each class of the classes is associated with a tumor; training a machine learning model to associate the nucleic acid data points of the training dataset with the identified classes; receiving the gene expression input data for classification; classifying, using the trained machine learning model, the gene expression input data as one or more of the identified classes; and outputting the classification of the gene expression input data.