Enterprise AI Analysis
AI-driven pre-screening for colorectal cancer using complete blood counts: toward broader population impact
An AI model using routine complete blood count (CBC) data for colorectal cancer (CRC) detection achieves an AUC of 0.77, offering a cost-effective and interpretable pre-screening tool.
Executive Impact
Early colorectal cancer (CRC) detection is critical, but traditional screening faces hurdles like limited colonoscopy access and low FIT adherence. This study introduces an AI model that uses routine CBC data for cost-effective CRC detection, aiming to improve risk stratification and resource allocation. The model, trained on RDW, SIRI, hemoglobin, and age, achieved an AUC of 0.77 for CRC, comparable to deep learning models, while remaining interpretable. It shows potential to augment existing screening programs.
0.77
CRC AUC
45-75
Target Age Range
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Key Findings
The AI model, leveraging RDW, SIRI, hemoglobin, and age, achieved an AUC of 0.77 for CRC detection, demonstrating interpretability and robustness. It outperformed traditional screening methods in resource-constrained settings by identifying high-risk individuals based on readily available CBC data.
0.77
CRC AUC
64%
Sensitivity
81%
Specificity
Enterprise Process Flow
Retrospective Data Collection (28,450 individuals, 45-75 yrs)
→
CBC & Colonoscopy Data (6-month window)
→
Data Labeling (CRC, Advanced Adenoma, Controls)
→
Dataset Splitting (70% training, 30% testing)
→
Feature Selection (RDW, SIRI, Hemoglobin, Age)
→
Ridge Regression Model Development
→
Model Performance Evaluation (AUC, Calibration, Interpretability)
Model Performance Comparison
| Feature | Ridge Regression Model | TabPFN (Deep Learning) |
|---|---|---|
| CRC AUC | 0.77 (95% CI: 0.75–0.77) | 0.77 (95% CI: 0.73–0.81) |
| Advanced Adenoma AUC | 0.60 (95% CI: 0.58–0.61) | 0.63 (95% CI: 0.61–0.65) |
| Interpretability | High (Transparent coefficients, SHAP values) | Low (Black-box, complex) |
| Stability | High (Robust performance) | Lower (Greater variability) |
| Computational Complexity | Low | High (Transformer-based, ensemble) |
16.48x
Higher CRC probability for high-risk individuals detected by AI model
The model identifies a high-risk group (3% of population) with a 16.48 times higher CRC probability than the average. This stratification allows for prioritized follow-up testing (FIT/colonoscopy) for those most in need, optimizing resource allocation in healthcare systems.
Case Study: Early Detection in Brazil
Challenge: Low adherence to existing screening methods (FIT) despite proven efficacy, leading to late-stage diagnoses.
Solution: Implement a CBC-based AI pre-screening tool to identify high-risk individuals, leveraging routine medical exams to improve accessibility and resource allocation for targeted follow-up.
Outcome: Enhanced early detection rates for CRC by improving adherence to subsequent diagnostic steps through risk-stratification, leading to better patient outcomes and reduced healthcare burden.
Calculate Your Potential ROI
Estimate the efficiency gains and cost savings AI could bring to your enterprise by streamlining diagnostic pre-screening workflows.
Estimated Annual Savings
$0
Annual Hours Reclaimed
0
Your AI Implementation Roadmap
A typical enterprise AI adoption journey. Each phase is tailored to your unique organizational structure and objectives.
Discovery & Strategy
Identify key challenges, define project scope, data readiness assessment, and establish success metrics.
Data Integration & Pre-processing
Securely integrate and cleanse relevant historical patient data, ensuring privacy compliance (e.g., LGPD).
Model Development & Customization
Develop or fine-tune AI models using your specific datasets for optimal performance in your clinical context.
Validation & Pilot Deployment
Rigorously test the AI solution in a controlled environment, gather feedback, and iterate for refinement.
Full-Scale Integration & Monitoring
Deploy the AI system into your existing diagnostic workflows, with continuous monitoring and optimization.
Ready to Transform Your Diagnostic Workflows?
Schedule a personalized consultation to explore how AI-driven pre-screening can enhance early detection and optimize resource allocation in your enterprise.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat