Evaluating the Compliance of MRI Brain Scans to American College of Radiology (ACR) Ordering Guidelines Using Text Mining
Technology
Training ＆ Education
Singapore Healthcare Management Congress
SingHealth
31 December 2018
To develop a predictive model using text mining to automate classification of MRI brain scans based on ACR guidelines. Automated text-mining tools streamline compliance assessment and can be adapted for broader applications.
Year Submitted: 2018
Published Date: 31 December 2018
Tags: Technology, Training ＆ Education, Digital Health, Data Analytics, Data Modelling, Education Research, Education Analytics
About this Content
Aims
To develop a predictive model using text mining to automate classification of MRI brain scans based on ACR guidelines.
Background
Manual classification of MRI indications was labor-intensive, necessitating automation to improve efficiency and accuracy.
Methods
Used anonymized patient reports, applied term frequency-inverse document frequency weighting, and tested algorithms like random forest, generalized linear model, and support vector machines.
Results
Random forest achieved the highest AUC (0.84), significantly reducing the manual workload for classification.
Conclusion
Automated text-mining tools streamline compliance assessment and can be adapted for broader applications.
Lessons Learnt
Random forest provides robust predictive capability, showing the value of machine learning in automating clinical processes.
Additional Information
Singapore Healthcare Management Congress 2018 – Digital Health Innovation
Keywords
Text Mining, MRI Compliance, ACR Guidelines
Innovators' Details
Innovators' Details
Healthcare Cluster(s) | SingHealth |
Organization(s) Involved | KK Women and Children Hospital |
Platform(s) | Singapore Healthcare Management Congress |
Healthcare Professional Group(s) | Allied Health |
Applicable Specialty or Discipline | Radiology |
Project Lead(s) | Tang PH |
Project Member(s) | Jing Xuan |
Connect with this contributor!
Tang PH - singaporehealthcaremanagement@singhealth.com.sg
Project Attachment
C_92_KKH_SHM_2018_Evaluating_Compliance_of_MRI_Brain_Scans_to_American_College_of_Radiology.pdf