A Review of the Challenges of Using Biomedical Big Data for Economic Evaluations of Precision Medicine

Abstract

There is potential value in incorporating biomedical big data (BBD)—observational real-world patient-level genomic and clinical data in multiple sub-populations—into economic evaluations of precision medicine. However, health economists face practical and methodological challenges when using BBD in this context. We conducted a literature review to identify and summarise these challenges. Relevant articles were identified in MEDLINE, EMBASE, EconLit, University of York Centre for Reviews and Dissemination and Cochrane Library from 2000 to 2018. Articles were included if they studied issues relevant to the interconnectedness of biomedical big data, precision medicine, and health economic evaluation. Nineteen articles were included in the review. Challenges identified related to data management, data quality and data analysis. The availability of large volumes of data from multiple sources, the need to conduct data linkages within an environment of opaque data access and sharing procedures, and other data management challenges are primarily practical and may not be long-term obstacles if procedures for data sharing and access are improved. However, the existence of missing data across linked datasets, the need to accommodate dynamic data, and other data quality and analysis challenges may require an evolution in economic evaluation methods. Health economists face challenges when using BBD in economic evaluations of technologies that facilitate precision medicine. Potential solutions to some of these challenges do, however, exist. Going forward, health economists who present work that uses BBD should document challenges and the solutions they have applied to the challenges to support future researcher endeavours. Electronic supplementary material The online version of this article (10.1007/s40258-019-00474-7) contains supplementary material, which is available to authorized users.

Machine Learning for Precision Health Economics and Outcomes Research (P-HEOR): Conceptual Review of Applications and Next Steps

“…While most applications of HEOR and CEA to precision medicine have focused on pharmacogenomics 37 to assess the value of precision testing to diagnose patients and align them with the appropriate targeted oncologic agent(s), 7 there is still a paucity of published research evaluating which precision patient subgroups can benefit from a particular cost-effective treatment regimen. 6 Some of the reasons are methodological and relate to our core understandings of bias and confounding. Individual treatment effect prediction could be hindered when predictive patient characteristics are also influencing treatment assignment 27 or when the scales for capturing a particular covariate are subject to measurement error, such as in mental health conditions.…”

“…6,7 This does not mean, however, that development work is not already being conducted to identify opportunities and foundational frameworks for the application of health economics tools to precision medicine, as suggested by Veenstra et al in Table 1. 7 As the economic evaluation of observational data and generation of real-world evidence will remain an important topic in precision medicine, 7 expert opinion suggests that contemporary advanced predictive algorithms such as machine learning (ML) should be explored as advanced tools for decision-making and cost-effectiveness analysis, 6 consistent with those particular domains outlined in Table 1. For example, when paired with clinical opinion from a health care professional, ML results can produce valuable insights into accelerating clinical workflow and optimizing therapeutic interventions and resource allocation.…”

“…For example, when paired with clinical opinion from a health care professional, ML results can produce valuable insights into accelerating clinical workflow and optimizing therapeutic interventions and resource allocation. 9 Aligned with recommendations to document challenges and potential solutions to support future developments in the application of health economics to precision medicine, 6 the objective of this article is to highlight the capabilities and limitations of ML applications in P-HEOR through a conceptualized example. We contextualize how a P-HEOR framework centered on recursive partitioning ML methods can be used to examine heterogeneity of treatment effect and to assess patient subgroups in which a particular intervention can be costeffective versus another.…”

“…Highlighting the aforementioned issues as possible reasons, a recent review identified no published economic evaluations using big data to inform precision medicine. 6 Literature reviews have also brought up the point that the application of HEOR to precision JOURNAL OF HEALTH ECONOMICS AND OUTCOMES RESEARCH Table 1.…”

Clinical Molecular Marker Testing Data Capture to Promote Precision Medicine Research Within the Cancer Research Network

“…14 These evidence gaps make it difficult to generate clinical practice guidelines and secure reimbursement for many molecular assays and tests, 15 and there is a need to incorporate real-world evidence to assess the effectiveness, value, and equitable use of these tests in clinical settings. [16][17][18][19] There are also challenges to implementing recommended precision medicine technologies in clinical settings and evaluating relevant outcomes. 20 Specifically, although there are clinical guidelines from professional societies about precision medicine, [21][22][23] many health care systems have not widely or comprehensively adopted standardized protocols governing when and how to use recommended genetic, molecular, and protein blood marker tests.…”