Medical Data Mining TRIZ Cases
- Time Series Deep Survival Analysis with Active Learning
Bidirectional LSTM networks extract temporal features from right-censored medical data to predict patient survival times.
- Real-Time Cardiovascular Outcome Prediction Using Dynamic Clinical Data
Sliding time windows update risk scores from electronic health records, providing 24-hour early warnings for cardiac arrest.
- Clear-Text Rule Engine for Healthcare Event Matching
A clear-text rule engine links patient healthcare events through a custom relational database extension.
- Mesh-Free Bone Model for Trabecular Fragmentation
Particle-based simulation resolves trabecular fragmentation inaccuracies, improving orthopedic implant stability predictions.
- EHR Agent Overlay for Patient Data Access
An EHR agent overlays patient information near a cursor within third-party applications, eliminating application switching.
- Automated Medical Data Validation via Machine Learning
Machine learning models validate medical test results, reducing manual review time and costs while maintaining accuracy.
- AI Diagnosis Assistant for Test Plan Optimization
Large language model proposes efficient test plans from patient symptoms.
- Impairment Rating Repair System Using Historical Data Correction
A system corrects incomplete impairment data sets using a high accuracy historical database to generate reliable temporary ratings.
- Orthodontic Treatment Planning with Reinforcement Learning
Automated analysis of dental arches identifies movement deviations to enable real-time plan adjustments and optimize treatment duration.
- Patient Monitoring Device Real-Time Behavior Analysis
A patient monitoring device classifies and correlates multi-source data to identify activity patterns.
- Autonomous Ear Infection Diagnosis via ML Biomarker Synthesis
A machine learning model synthesizes patient history and ear biomarkers to generate probability-based diagnoses.
- Blood Glucose Probability Prediction Using Kernel Density Estimation
Kernel density estimation with separate bandwidths predicts adverse blood glucose probabilities, resolving sparse spot monitoring data limitations.
- Stress Detection via Resting-State Physiological Baselines
A wearable system establishes individual physiological baselines during rest to determine stress levels without requiring stressful calibration states.
- Healthcare Analytics Management System with Closed-Loop Feedback
A healthcare analytics management system deploys models and feeds performance deviations back to development modules for continuous improvement.
- Question Generation System for Automated Medical Diagnosis
A question generation system uses a knowledge graph to identify candidate symptoms and generate targeted questions.
- Unified Health Platform Integrating Patient Data
A unified health platform organizes fragmented patient records into a single view for authorized providers.
- EEG Brain Region Interaction Analysis for Psychiatric Cohort Identification
Scalp electroencephalograph data analyzes brain region mutual interaction characteristics to identify psychiatric disorder patient cohorts.
- Adaptive Time-Frequency Windows for iEEG Seizure Onset Zone Localization
Computing system filters intracranial electroencephalography signals within dynamic time-frequency windows to extract feature values.
- Wearable Food Intake Detection via Modular Signal Processing
Modular wearable system extracts physiological features to detect food intake events autonomously.
- Knowledge Graph Embedding Prediction for New Node Links
Information processing device specifies similar graph data from training sets to predict authenticity for new node links without re-training entire datasets.
- Data Processing System for Unstructured Source Data
Segmented modules translate unstructured data to boost processing speed while maintaining manageable system complexity.
- Media Rendering Synchronization with Breathing Phases
A processing system aligns media content rendering timing with predicted user breathing phases to enhance attention and comprehension.
- Health Risk Determination Using Multi-Source Data Integration
Multi-source machine learning models analyze patient, environmental, and behavioral data to determine health risks without frequent doctor visits.
- Health Trajectory Fingerprinting for Clinical Efficacy
A computing system generates digital fingerprints from segmented health trajectories to identify similar patient sub-trajectories.
- Diagnostic Effectiveness Tool for Probabilistic Test Selection
A computing system generates probabilistic suggestions for diagnostic tests using validated medical knowledge models.
- Diagnoses-Based Disease Prediction Module
A diagnoses-based prediction module applies machine learning to population and subject-specific disease data.
- CNN Blood Glucose Prediction Model Personalization
A convolutional neural network processes image-based glucose data to predict blood glucose conditions.
- Neural Network Weight Update via Augmented Image Difference
A neural network training method generates augmented images and updates weights based on output differences.
- PCG Signal Quality Classification via mRMR Feature Ranking
A random forest classifier evaluates phonocardiogram signals using minimum redundancy maximum relevance feature extraction to determine recording quality.
- Multimodal Cardiac Prediction via Unified Deep Learning Architecture
A single architecture deep learning model processes electrocardiogram and clinical data through isolated layers to generate normalized outputs.
- Clinical Risk Model Bias Calibration
A model-assisted system analyzes medical records to determine patient risk levels.
- Augmented Reality Appetite Enhancement System
An augmented reality system stimulates user senses through visual and auditory stimuli to enhance appetite.
- ATAC-seq Data Normalization via Control Peak Selection
A method normalizes ATAC-seq data using overlapping peaks and DNase I hypersensitivity consensus regions to enable quantitative comparison across samples.
- MALDI-TOF Mass Spectrometry for Early HCC Detection
MALDI-TOF mass spectrometry analyzes serum protein patterns to identify early-stage hepatocellular carcinoma in high-risk patients.
- ML Classification for Sepsis Detection Using Routine Blood Tests
Machine learning classification system analyzes routine blood test data to identify disease risk.
- False Alarm Detection in Patient Monitoring via Artifact Feature Classification
An ensemble classifier extracts artifact features from alarm signals to distinguish true positives from false positives.
- Medical Device Lending Prediction System
A prediction system uses learned models to forecast medical device return times based on electronic chart data and lending records.
- Real-Time PPG Signal Quality Assessment via Frequency Domain Analysis
A system classifies photoplethysmogram signal quality using frequency domain analysis and differential evolutionary optimization.
- Multi-Sensor Health Platform with ML-Powered Energy Scheduling
A multi-sensor health platform uses machine learning to predict patient activity and schedule sensor recharging cycles.
- Electronic System-MD for Psychiatric Medication Assessment
CelestHealth System-MD electronically monitors psychiatric medication effectiveness using Behavioral Health Measure-20 questionnaires.
- Machine Learning Diagnosis Models Data Integration
Machine learning models standardize disparate medical records to improve diagnostic accuracy while managing computational complexity.
- Medical Imaging Parameter Recommendation System for Radiology Training
A computer system generates training materials from reference imaging examinations to display suggested scan parameters on a user interface.
- Machine Learning Model Training via Shared Parameters
A machine learning training method shares model parameters between distinct sample sets to enhance accuracy without exchanging raw data.
- Predictive Maintenance Sequence Model for Medical Imaging Root Cause Analysis
A sequence model analyzes feature values to generate specific service actions for medical imaging component failures.