Patient-specific meta-analysis for risk assessment using multivariate proportional hazards regression

Area of Science:

• Biostatistics
• Clinical Epidemiology
• Health Outcomes Research

Background:

• Accurate prediction of future clinical events is crucial for personalized medicine.
• Existing methods often struggle to integrate data from multiple clinical trials or cohort studies effectively.
• Meta-analysis techniques offer a powerful framework for synthesizing evidence but require adaptation for individual patient risk prediction.

Purpose of the Study:

• To develop and validate a novel meta-analysis method for assessing individual patient risk of future clinical events.
• To enable the integration of data from diverse clinical trial and cohort studies.
• To provide a flexible framework applicable to various regression models.

Main Methods:

• Proposed a method combining patient-specific log cumulative hazard estimates across studies.
• Utilized Cox proportional hazards regression and meta-analysis techniques (fixed- and random-effects models).
• Employed weighting by the relative precision of estimates from each study.

Main Results:

• The method successfully combines information from multiple studies for risk prediction.
• Patient risk assessment can be performed using summary statistics derived from each study.
• Simulations and real-data applications demonstrated the method's validity and utility.
• The approach is generalizable to logistic regression and linear models.

Conclusions:

• The developed meta-analysis method provides a robust approach for individual patient risk prediction.
• This technique enhances the utility of existing clinical trial and cohort data for future event forecasting.
• The method offers a significant advancement in synthesizing multi-study data for clinical decision-making.