Using Similarity Metrics on Real World Data and Patient Treatment Pathways to Recommend the Next Treatment.

Area of Science:

• Oncology
• Biostatistics
• Health Informatics

Background:

• Non-small-cell lung cancer (NSCLC) has a poor five-year survival rate.
• Current NSCLC treatments are often population-based, not individualized.
• Identifying optimal therapies for individual NSCLC patients remains a challenge.

Purpose of the Study:

• To develop an analytical approach for recommending the next best treatment for NSCLC patients.
• To assess if patient similarity metrics and treatment history can guide personalized therapy selection.
• To evaluate the current standard of care against data-driven treatment recommendations.

Main Methods:

• Utilized the Gower similarity metric to quantify patient similarity.
• Integrated prior treatment knowledge with patient similarity data.
• Conducted a retrospective analysis of NSCLC patient treatment data.
• Employed patient analytics to identify potential treatment recommendations.

Main Results:

• A significant proportion of NSCLC patients were not recommended the therapy associated with the best survival outcomes upon requiring a new treatment.
• Patient analytics demonstrated potential to complement clinical decision-making in NSCLC treatment selection.
• The study highlights a gap between available therapies and optimal patient-specific treatment pathways.

Conclusions:

• Analytical methods, incorporating patient similarity and treatment history, can aid in recommending optimal NSCLC therapies.
• Current treatment selection processes may not consistently identify the most effective therapy for individual patients.
• Further research is needed to validate these findings and explore patient outcomes beyond survival.