A supervised machine learning model for identifying predictive factors for recommending head and neck cancer surgery
Max L Jiam1, Kevin Z Xin2, Patrick K Ha3
1School of Computer Science, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA.
Head & Neck
|February 12, 2024
Summary
Machine learning accurately predicts head and neck cancer surgery, potentially reducing treatment delays. This AI tool can improve patient referral processing for faster care.
Area of Science:
- Oncology
- Data Science
- Health Informatics
Background:
- Patient referrals are often handled by non-medical staff, leading to potential delays.
- Incorrect referral processing in oncology can have severe consequences for treatment timelines.
- Identifying factors for surgical resection can optimize head and neck cancer patient referrals.
Purpose of the Study:
- To develop a supervised machine learning (ML) platform.
- Identify variables associated with head and neck surgical resection likelihood.
- Improve patient referral processing and expedite cancer treatment.
Main Methods:
- Retrospective cohort study design.
- Utilized 64,222 patient data points from the SEER database.
- Developed and applied a random forest ML model.
Main Results:
- The random forest ML model achieved 81% accuracy in classifying patients offered head and neck surgery.
- Model demonstrated 86% sensitivity and 71% specificity.
- Positive and negative predictive values were 85% and 73%, respectively.
Conclusions:
- Machine learning accurately predicts head and neck cancer surgery recommendations.
- ML models can utilize patient and cancer data from large datasets.
- ML integration into referral processes may decrease patient treatment times.
