基于机器学习的临床质谱的风险预测方法及设备
By constructing a 25-dimensional feature space and a hybrid ensemble learning model for the LC-MS system, the problems of lag and rigidity of the LC-MS system are solved, enabling forward-looking risk warning for the mass spectrometry system and improving the accuracy and efficiency of detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI CLINICAL LAB CENT
- Filing Date
- 2026-04-27
- Publication Date
- 2026-07-17
AI Technical Summary
Existing LC-MS mass spectrometry systems suffer from lag, bias, and rigidity during detection, leading to frequent false positive alarms and an inability to effectively handle complex nonlinear coupling relationships across multiple feature dimensions, thus affecting the accuracy and efficiency of detection results.
A machine learning-based approach is used to construct a 25-dimensional feature space. Random forest and XGBoost models are used for risk prediction. By fusing multi-dimensional features and adjusting dynamic thresholds, the nonlinear coupling relationship of the system is captured, enabling proactive early warning.
It significantly improves the accuracy and predictability of early warning in mass spectrometry systems, reduces false alarm rates, and realizes the shift from post-event error correction to pre-event prevention, thereby reducing sample waste and detection delays.
Smart Images
Figure CN122084815B_ABST