# The value for money of artificial intelligence-empowered precision medicine: a systematic review and regression analysis

## TL;DR

This study reviews the cost-effectiveness of AI in precision medicine, finding it mostly cost-effective but with significant variability.

## Contribution

The study introduces a regression analysis with machine learning to explore value heterogeneity in AI-empowered precision medicine.

## Key findings

- AI-PM was cost-saving or cost-effective in 89% of base-case analyses.
- Incremental cost-effectiveness ratios ranged from dominant to $129,174/QALY.
- Modeling choices and system-level factors caused significant value heterogeneity.

## Abstract

Artificial intelligence has empowered precision medicine (AI-PM) to transform healthcare. This study synthesized available evidence on the cost-effectiveness of AI-PM. We systematically searched five major databases for economic evaluations of AI-PM, extracted data, and assessed risk-of-bias using the Bias in Economic Evaluation (ECOBIAS) checklist. For cost-utility analyses, the value-for-money was quantitatively summarized, and regression analyses incorporating machine learning were conducted to explore value heterogeneity. Forty-eight economic evaluations were included, of which 31 were cost-utility analyses. Although risk-of-bias assessment indicated potential systematic optimism, AI-PM was cost-saving or cost-effective in 89% of base-case analyses, with incremental cost-effectiveness ratios ranging from dominant to $129,174/quality-adjusted life-year (QALY). Interquartile ranges of incremental costs (−$259 to $28), QALY gains (0.001–0.019), and net monetary benefits (NMB; $18 to $986 at a willingness-to-pay threshold equal to one-time per-capita GDP) indicated modest health gains at minimal additional costs, and likely high value heterogeneity. Modeling choices and system-level factors were identified as essential sources of heterogeneity in estimated NMBs. Additional value assessment revealed low adaptability and underreported key value factors, leaving significant uncertainties in AI-PM adoption.