Predicting Long-Term Self-Rated Health in Small Areas Using Ordinal Regression and Microsimulation
Seán Caulfield Curley, Karl Mason, Patrick Mannion · 2026
Read Predicting Long-Term Self-Rated Health in Small Areas Using Ordinal Regression and Microsimulation by Seán Caulfield Curley, Karl Mason, Patrick Mannion (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-14335.txt.
Introduction
Seán Caulfield Curley, Karl Mason, and Patrick Mannion forecast local health in this 2026 paper. Their target is self-rated health, how people judge their own health, in small areas. They predict it for a future population, not just the present one. Each person's socio-economic traits feed the forecast.
The team uses an open-source microsimulation that projects Ireland's population into the future. In the simulation, each person carries demographic and socio-economic traits. Ordinal regression then maps those traits to health ratings, which come in ordered grades. The aim is long-term health estimates for small areas, the scale at which local services plan.
Catalog record
| CREATOR | Seán Caulfield Curley, Karl Mason, Patrick Mannion |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 55 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-14335
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Predicting Long-Term Self-Rated Health in Small Areas Using Ordinal Regression and Microsimulation Seán Caulfield Curley, Karl Mason, Patrick Mannion Abstract This paper presents an approach for predicting the self-rated health of individuals in a future population utilising the individuals’ socio-economic characteristics. An open-source microsimulation is used to project Ireland’s population into the future where each individual is defined by a number of demographic and socio-economic characteristics. The model is disaggregated spatially at the Electoral Division level, allowing for analysis of results at that, or any broader geographical scales. Ordinal regression is utilised to predict an individual’s self-rated health based on their socio-economic characteristics and this method is shown to match well to Ireland’s 2022 distribution of health statuses. Due to differences in the health status distributions of the health microdata and the national data, an alignment technique is proposed to bring predictions closer to real values. It is illustrated for one potential future population that the effects of an ageing population may outweigh other improvements in socio-economic outcomes to disimprove Ireland’s mean self-rated health slightly. Health modelling at this kind of granular scale could offer local authorities a chance to predict and combat health issues which may arise in their local populations in the future. Keywords: Microsimulation, Ordinal Regression, Self-Rated Health, Small Area Health, Ireland 1 Introduction Accurate anticipation of the health trajectories of future populations is central to informing public health planning, healthcare resource allocation, and policy design. Self-rated health (SRH), a widely used measure that captures an individual’s overall perception of their health status, has been shown to predict morbidity ( Chandola & Jenkinson 2000 ) , healthcare utilization ( Hansen et al. 2002 ; Chamberlain et al. 2014 ) , and mortality ( Lorem et al. 2020 ) . Its multidimensional nature, integrating both physical and psychosocial aspects of health, makes SRH a valuable outcome for population health forecasting. Previous approaches using microsimulations to model health outcomes have largely either not focused on the temporal aspect of health (analysing the current population only) or relied on aggregate-level projections. In effect, the method proposed in this paper aims to combine these previous approaches to predict future health outcomes while incorporating the heterogeneity of health with respect to geography. Figure 1: A graphical summary of how the approach operates. Note that synthetic individuals are created for this example and are not based on individuals from either dataset. In this study, we integrate microsimulation with ordinal regression modelling to predict future distributions of self-rated health. Ordinal regression is well-suited for SRH, which is typically measured on an ordered categorical scale (e.g. “Very Good” to “Very Bad”), allowing us to model the probability of individuals reporting different health levels based on their demographic profiles. By combining this regression framework with a microsimulation model, we may generate forward-looking estimates of population health under various demographic scenarios. A graphical summary of the approach is given in Figure 1 . This work contributes to the field of health forecasting by providing a methodological approach that combines the strengths of microsimulation and ordinal regression. The resulting projections can support policymakers and public health professionals in anticipating future health needs, identifying vulnerable groups, and designing interventions that are responsive to demographic change. “What-if” scenarios can also be tested, both with regard to the microsimulation and the regression, meaning alternate scenarios than the ones proposed in this paper can be tested easily. The rest of this paper is laid out as follows: Section 2 summarises the background to the problem, giving the motivation for using SRH, highlighting some relevant Irish microsimulations, and illustrating how microsimulations have been used in a health context previously. The next section goes in-depth in the workings of the approach. Each module of the microsimulation is summarised and the operation of ordinal regression is explained. The data used to support both methods is also described in detail. A method to align the health microdata with the observed Census data is introduced in subsections 3.3 and 3.4 . Section 4 presents the results for the approach, both in terms of validation and results of future projections. After this, a concise example application of the resulting populations is outlined in Section 4.6 . The paper concludes with a discussion on some of the strengths and limitations of the approach. 2 Background Microsimulation is a simulation paradigm in which simulation units represent low-level entities such as individual people, households, or farms. Dynamic microsimulations, where the temporal effect on the topic in question is analysed, are particularly powerful for analysing demographic changes at the individual level, and have been used to model population change in Canada ( Morency et al. 2025 ) , Germany ( Münnich et al. 2021 ) and the US ( The Urban Institute 2015 ) . One of the key outputs of microsimulation is enabling Agent-Based Modelling (ABM), where individuals in the population are treated as agents who can interact with each other and the environment. One example of an agent-based model being adapted from microsimulations is a model of urban freight distribution where microsimulation models passenger trips, and freight entities like customers, carriers and shippers are treated as agents ( Gómez-Marín et al. 2018 ) . An example from demography is the model introduced by Bae et al. 2016 , which handles population size drivers such as fertility, mortality and migration with microsimulation and uses ABM to model the process of an individual finding a partner for marriage. There have been multiple microsimulation studies published with Ireland as the focus. The Economic and Social Research Institute (ESRI) used survey and register data to model Ireland’s tax-benefit system with a microsimulation ( Keane et al. 2023 ) . The Irish Farming Land Rental Market has also been modelled using the technique ( Loughrey & Hennessy 2022 ) . O’Donoghue et al. 2013 pioneered microsimulation in Ireland by using a static model to create a synthetic population for the country. A dynamic, open-source model inspired by that model and those of the US, Canada and Germany was also recently released ( Caulfield Curley et al. 2025a ) . A large contributing factor to the breadth of research performed to date in Ireland is the wealth of data provided by the Central Statistics Office (CSO). Data featured on the data.cso.ie repository includes Census data, Small Area Population Statistics (SAPS), and the results of the CSO’s population projections. The CSO publish a report a few years after each Census outlining the results of their population projections ( CSO 2023 ) which are carried out using the Demographic Component Method (DCM). Their most recent report projects the Irish population from 2023 to 2057, disaggregated at the NUTS3 region level 1 1 1 The Nomenclature of Territorial Units for Statistics (NUTS) are a standard for referencing country subdivisions. In Ireland, the NUTS3 regions are Border, West, Mid-West, Mid-East, South-East, South-West, Dublin and Midlands . There have also been numerous applications of microsimulation to tackle health-related issues. Wu et al. 2022 used static microsimulation techniques (without a temporal component) to create a synthetic population dataset relating health and socio-economic factors for small areas in Great Britain. A similar approach was used to generate a population which included both personal