An intelligent quality control system for High-Performance Liquid Chromatography (HPLC) that automatically detects anomalous chromatograms using machine learning and rule-based diagnostics. The project combines global signal normalization, Isolation Forest anomaly detection, and chromatographic quality checks to assist laboratory quality control workflows.
High-Performance Liquid Chromatography (HPLC) is widely used in pharmaceutical, clinical, food, and environmental laboratories for compound separation and quality assurance. Instrument failures, baseline drift, peak distortion, and unexpected chromatographic behaviour can compromise analytical results and often require manual inspection.
This project presents an automated HPLC quality control workflow that analyzes chromatographic signals, identifies abnormal instrument behaviour using machine learning, and provides interpretable diagnostic messages through rule-based quality assessment.
The system demonstrates how machine learning can support laboratory quality control while maintaining transparent and explainable decision-making.
- Project Overview
- Objectives
- Features
- Repository Structure
- Methodology
- Machine Learning Workflow
- Results
- Technologies Used
- Future Improvements
- Acknowledgements
- License
The HPLC QC System processes chromatographic signals through a sequence of preprocessing, anomaly detection, and diagnostic steps.
The workflow includes:
- Chromatogram preprocessing
- Global normalization
- Feature extraction
- Isolation Forest anomaly detection
- Rule-based diagnosis
- Quality assessment report generation
- Detect abnormal chromatograms automatically.
- Reduce manual inspection of HPLC runs.
- Provide explainable quality control decisions.
- Improve anomaly detection using machine learning.
- Demonstrate a reproducible QC workflow.
- Automated chromatogram preprocessing
- Global normalization
- Isolation Forest anomaly detection
- Rule-based diagnosis
- Interactive quality assessment
- Performance evaluation using classification metrics
HPLC-qc-system/
│
├── app/
├── data/
├── docs/
├── figures/
├── notebooks/
├── src/
├── README.md
├── requirements.txt
├── LICENSE
└── .gitignore
| Tool | Purpose |
|---|---|
| Python | Core programming language |
| NumPy | Numerical computing |
| Pandas | Data processing |
| Matplotlib | Data visualization |
| Scikit-learn | Machine learning |
| Isolation Forest | Anomaly detection |
| Streamlit | Interactive web application |
| Software | Version |
|---|---|
| Python | 3.11+ |
| NumPy | Latest compatible |
| Pandas | Latest compatible |
| Matplotlib | Latest compatible |
| Scikit-learn | Latest compatible |
| Streamlit | Latest compatible |
The HPLC quality control workflow consists of the following stages:
- Import chromatographic data.
- Apply global normalization.
- Extract relevant signal characteristics.
- Detect anomalous chromatograms using Isolation Forest.
- Perform rule-based diagnostic analysis.
- Generate quality assessment results.
Chromatogram
│
▼
Global Normalization
│
▼
Feature Extraction
│
▼
Isolation Forest
│
▼
Rule-Based Diagnosis
│
▼
Quality Assessment
The implemented workflow successfully demonstrated automated anomaly detection for HPLC quality control.
| Metric | Value |
|---|---|
| Baseline F1 Score | 0.61 |
| Final Validation F1 Score | 0.87 |
The improvement in validation performance demonstrates the effectiveness of combining global normalization with Isolation Forest–based anomaly detection.
- Machine Learning
- Anomaly Detection
- Scientific Data Analysis
- Chromatographic Data Processing
- Python Programming
- Data Visualization
- Quality Control Automation
- Real-time HPLC monitoring
- Deep learning–based anomaly detection
- Multi-instrument support
- Cloud deployment
- Laboratory Information Management System (LIMS) integration
This project was developed as a machine learning application for automated HPLC quality control and chromatographic anomaly detection. It demonstrates the integration of explainable machine learning techniques with laboratory quality assurance workflows.
This project is licensed under the MIT License.
Shreejan V
B.Tech Biotechnology (Computational Biology)
SRM Institute of Science and Technology
GitHub: https://github.com/shreejan-v