Upcoming Planned Work:

  • Finalize pipeline integration and deployment into clinical tools
  • Implementation of model pipelines into the clinic environment


MIRA Clinical Learning Environment (MIRA-CLE) for Lung

Co-PIs: Andrew Hope, Tony Tadic, Geoff Liu

Machine learning models can provide clinicians with additional insights about their patients, supporting clinical decision making at the point of care. Clinicians have access to all the same information they would normally possess but with the addition of specific information summarized from modeling hundreds or thousands of other similar cases. This easy access to comparison data and summarized patient information can provide added insight to clinicians about individual patients.

As part of MIRACLE, machine learning (ML) models have been created and tested along three aims, to highlight lung cancer patients who may be at higher risk of (1) inflammatory lung disease (ILD), (2) local failures, distant metastases and reduced overall survival based on tumour specific growth rate (SGR), and (3) treatment toxicities based on comparisons of cone beam CT (CBCT) images over time.

Being able to proactively identify patients who are prone to these risks through subtle clues available in a patient’s image, or by comparing and measuring features of a patient’s images over time, and then highlighting this information for radiation oncologists is expected to improve the care provided to our patients, as well as patient outcomes and patient safety.

Key Milestones:

  • Model development, testing and validation completed or nearly completed for all three aims
  • Pipeline integration and deployment into clinical tools in progress for all three aims
  • Web application created for researcher/clinician access to pipelines
  • Quality Improvement (QI) application for clinical implementation submitted