Assessing clinician performance using a multi-modality clinical decision-support system for lung cancer...

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Surgery is the primary treatment for early-stage lung cancer.
• Identifying patients who benefit from adjuvant therapy post-surgery is challenging.
• Accurate prognostication is crucial for personalized lung cancer treatment.

Purpose of the Study:

• To develop a clinical decision support system (CDSS) for post-surgery lung cancer prognostication.
• To integrate a multi-modality deep learning model (DLM) into the CDSS.
• To evaluate the impact of the CDSS on clinician decision-making and confidence.

Main Methods:

• A multi-modality deep learning model (DLM) was developed using pre-operative medical images and clinical, surgical, and pathological data.
• The DLM was externally validated.
• A CDSS was created to present patient information and DLM results to clinicians.
• Four oncologists assessed patient recurrence risk and recommendations with and without CDSS-DLM information.

Main Results:

• The CDSS, incorporating DLM insights, showed potential to improve clinicians' prediction performance.
• Clinician confidence in prognostication and treatment recommendations was enhanced by the CDSS-DLM information.
• This study represents the first integration of a multi-modality DLM for prognostication within a CDSS for lung cancer.

Conclusions:

• The developed CDSS with an integrated multi-modality DLM shows promise for improving post-surgery lung cancer prognostication.
• Clinician evaluation suggests the CDSS can enhance prediction accuracy and confidence.
• This work pioneers the use of such integrated systems and explores clinician acceptance in lung cancer care.