Automated Excel-to-R Conversion of Health Economic Models Using the REEEVR Tool: Results of Alpha Test
Author(s)
Michael J. O'Donnell, PhD1, Qian Xin, MSc2, Abdul-Lateef Haji-Ali, Ph.D3, Gabriel Rogers, MSc4, Howard Thom, BA, MSc, PhD5.
1Senior Research Associate, University of Bristol, Bristol, United Kingdom, 2University of Bristol, Oxford, United Kingdom, 3Heriot Watt, Edinburgh, United Kingdom, 4University of Manchester, Manchester, United Kingdom, 5Bristol Medical School : Population Health Sciences, Bristol, United Kingdom.
1Senior Research Associate, University of Bristol, Bristol, United Kingdom, 2University of Bristol, Oxford, United Kingdom, 3Heriot Watt, Edinburgh, United Kingdom, 4University of Manchester, Manchester, United Kingdom, 5Bristol Medical School : Population Health Sciences, Bristol, United Kingdom.
OBJECTIVES: Perform alpha test of the Reliable and Efficient Estimation of the Economic Value of medical Research (REEEVR) Excel-to-R conversion software on models developed by academia, industry consulting and HTA agencies.
METHODS: REEEVR performs automated Excel-to-R conversion by parsing an input Excel file, generating abstract syntax trees for each cell, converting the cell into R, then reconstructing the program fully in R. We recently finished a closed alpha trial, where we invited potential users from across academia and industry to evaluate the software and submit feedback. As part of this process, extensive logging was performed, with these logs being requested from each participant to allow for further development of the REEEVR tool. To achieve a wide variety of users, the use of both informal networks and social media via LinkedIn were used.
RESULTS: We sent invitations to 24 people, of which 9 responded with the requisite logs and feedback. Of those that responded, all participants tested the tool on multiple Excel documents, providing 20 log files for a variety of Excel models. Of those models tested, only one model successfully converted and ran. Reasons for unsuccessful conversion fall into three categories: Missing function implementations, custom function implementations, unexpected but consistent crash of the tool. We identified 32 unique functions which, if implemented, would likely allow successful conversion of most models included in the alpha test.
CONCLUSIONS: We found significant interest in the social media space and informal networks for this tool, however, participation in the study is limited to those with time and inclination to work with a tool with known and unknown issues. The feedback from the alpha test has allowed us to pivot development, which we expect will allow a beta test to be performed in November of 2025, which we predict will have significantly better conversion ratios.
METHODS: REEEVR performs automated Excel-to-R conversion by parsing an input Excel file, generating abstract syntax trees for each cell, converting the cell into R, then reconstructing the program fully in R. We recently finished a closed alpha trial, where we invited potential users from across academia and industry to evaluate the software and submit feedback. As part of this process, extensive logging was performed, with these logs being requested from each participant to allow for further development of the REEEVR tool. To achieve a wide variety of users, the use of both informal networks and social media via LinkedIn were used.
RESULTS: We sent invitations to 24 people, of which 9 responded with the requisite logs and feedback. Of those that responded, all participants tested the tool on multiple Excel documents, providing 20 log files for a variety of Excel models. Of those models tested, only one model successfully converted and ran. Reasons for unsuccessful conversion fall into three categories: Missing function implementations, custom function implementations, unexpected but consistent crash of the tool. We identified 32 unique functions which, if implemented, would likely allow successful conversion of most models included in the alpha test.
CONCLUSIONS: We found significant interest in the social media space and informal networks for this tool, however, participation in the study is limited to those with time and inclination to work with a tool with known and unknown issues. The feedback from the alpha test has allowed us to pivot development, which we expect will allow a beta test to be performed in November of 2025, which we predict will have significantly better conversion ratios.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
HTA50
Topic
Economic Evaluation, Health Technology Assessment, Study Approaches
Disease
No Additional Disease & Conditions/Specialized Treatment Areas