Moderate evidence
More
AI cost model in Bergamo, Italy, systematically underestimated the highest-cost patients
Study title: An application of random forest regression for predicting healthcare costs using administrative databases
- Evidence type
- Observational study
- Sources
- 1 sourcePeer-reviewed
- Published
- 1 min read
- Truvace record
- TRV-2026-1163cite or check it
Study title: An application of random forest regression for predicting healthcare costs using administrative databases
Background Accurate healthcare cost predictions are essential for health economic decision-making. We propose a bottom-up approach that aggregates individual-level predictions to estimate expenditures for specific population segments without relying on a priori segmentation.
Read the study abstract
Background Accurate healthcare cost predictions are essential for health economic decision-making. We propose a bottom-up approach that aggregates individual-level predictions to estimate expenditures for specific population segments without relying on a priori segmentation. Methods This retrospective population-based study used administrative healthcare data (2011-2023) from the Health Protection Agency of Bergamo (Italy). We traced 5-year healthcare resource utilization to predict individual's healthcare costs…
The problem
In a retrospective study of Bergamo, Italy health data, random forest models predicting individual healthcare costs performed poorly and systematically underestimated high-cost users.
In brief
We traced 5-year healthcare resource utilization to predict individual's healthcare costs in the following year using random forest regression algorithms. Results At the individual-level, algorithms had a poor performance, systematically underestimating high-cost users.
Main points
- Background Accurate healthcare cost predictions are essential for health economic decision-making.
- We propose a bottom-up approach that aggregates individual-level predictions to estimate expenditures for specific population segments without relying on a priori segmentation.
- Methods This retrospective population-based study used administrative healthcare data (2011-2023) from the Health Protection Agency of Bergamo (Italy).
The full story
Background Accurate healthcare cost predictions are essential for health economic decision-making. We propose a bottom-up approach that aggregates individual-level predictions to estimate expenditures for specific population segments without relying on a priori segmentation.
Methods This retrospective population-based study used administrative healthcare data (2011-2023) from the Health Protection Agency of Bergamo (Italy). We traced 5-year healthcare resource utilization to predict individual's healthcare costs in the following year using random forest regression algorithms.
Sources
1 on file- Expert Review of Pharmacoeconomics & Outcomes Researchdoi:10.1080/14737167.2026.2738028
Talk it over
The random forest models systematically underestimated high-cost users in Bergamo. Who would carry the cost if a health agency budgeted using these predictions?
AI cost model in Bergamo, Italy, systematically underestimated the highest-cost patientshttps://truvace.com/item/an-application-of-random-forest-regression-for-predicting-healthcare-costs-using
The random forest models systematically underestimated high-cost users in Bergamo. Who would carry the cost if a health agency budgeted using these predictions?
Answer in the threadThe models learned from administrative data from the Health Protection Agency of Bergamo, which records resource use but not why people needed care. How might that limit the predictions for high-cost users, and what would settle it?
Answer in the threadThe researchers propose adding up individual predictions to estimate spending for population segments. What would you want to see before trusting that approach, given the poor individual-level results?
Answer in the thread
If you have more time
Look up how the segment-level estimates compared with actual spending. Which result would settle whether the poor individual-level performance matters?
Answer in the threadIf a regional health service you know used a cost model like this, who should check it and how often?
Answer in the thread
ace
What readers say