Artificial Intelligence Algorithms in Predictive Factors for Hematologic Toxicities During Concurrent Chemoradiation
Ion Petre1,2, Serban Negru3,4, Radu Dragomir5
1Department of Biostatistics, Victor Babes University of Medicine and Pharmacy, Timisoara, ROU.
Cureus
|November 4, 2024
Summary
This study evaluated chemotherapy toxicities in cervical cancer (CC) patients, finding that AI models accurately predict CC stages. Early detection of cervical cancer significantly improves patient prognosis and quality of life.
Area of Science:
- Oncology
- Medical Informatics
Background:
- Cervical cancer (CC) treatment varies by stage, with similar survival rates for several therapies.
- Hematologic toxicities from chemotherapy can negatively impact patient quality of life.
Purpose of the Study:
- To determine hematologic toxicities during concurrent chemotherapy for cervical cancer using cisplatin or carboplatin.
- To develop and evaluate artificial intelligence (AI) models for predicting cervical cancer stages.
Main Methods:
- Hematologic markers were analyzed using statistical models.
- AI models, including Naïve Bayes, Random Forest, Decision Trees, and TabPFN, were trained on patient data to forecast CC stage.
- Model performance was assessed using classification scores, accuracy, and computational complexity.
Main Results:
- Naïve Bayes, Random Forest, and Decision Trees achieved 100% accuracy in predicting cervical cancer stages.
- TabPFN demonstrated 88% accuracy.
- Hematologic toxicities were observed to increase linearly with decreasing hematologic markers.
Conclusions:
- AI models show high potential for accurate cervical cancer staging.
- Early detection of cervical cancer is crucial for improving prognosis and patient outcomes.
- Understanding and managing chemotherapy-induced hematologic toxicities is vital for enhancing patient quality of life.
