1. Limitations of national-level indicators in heterogeneous settings

Country-level indices such as the Human Development Index (HDI), sanitation coverage, and gross national income per capita are useful initial proxies for infection risk. However, these aggregated metrics often mask substantial subnational disparities. For instance, although Thailand boasts over 90% national sanitation coverage, marginalized communities along the Thai–Myanmar border continue to face poor sanitation and limited healthcare access [2,3]. Similar patterns exist in Indonesia and the Philippines [4]. Relying solely on national averages risks underestimating true exposure in vulnerable subpopulations. Evidence from Peru similarly shows poor correlation between regional HDI and parasitic infection prevalence, underscoring the need for more granular, district- or community-level data to enhance model accuracy [5].

Additional disparities are evident across other Southeast Asian nations that were not discussed in the original framework by Purkiss et al. In Lao People’s Democratic Republic (Lao PDR) and Vietnam, despite progress in health infrastructure, community-based surveys continue to report endemic levels of STH infections in remote and mountainous areas, particularly among school-aged children and ethnic minority groups. For example, a WHO-supported study found that over 23% of children in northern Lao PDR tested positive for Ascaris lumbricoides or Trichuris trichiura, despite ongoing deworming campaigns [6].

In Malaysia, while national health indicators appear strong, localized studies have identified persistent STH transmission among indigenous Orang Asli populations and undocumented migrants working in agriculture. A recent study found STH prevalence exceeding 50% in some Orang Asli communities, associated with poor sanitation and lack of health education [7]. Similarly, in Timor-Leste, national-level data mask continued high STH prevalence in rural districts, with reports of over 30% prevalence in school-aged children in Ermera and Bobonaro provinces [8].

Conversely, Singapore and Brunei Darussalam, though classified as low-endemic, host sizeable migrant worker populations from STH-endemic countries. These groups may still be at risk due to overcrowded housing, poor sanitation in dormitories, and limited access to deworming programs [9].

These examples reinforce the need for predictive models to integrate subnational data and vulnerability factors to better reflect real-world risk.

2. Post-migration exposure and contextual risk factors

Purkiss et al.’s model primarily addresses pre-migration exposures, assuming infection originates in the migrant’s country of origin. Our observations among migrants from Myanmar and Cambodia in Thailand, however, suggest ongoing exposure risks post-migration. Many migrants reside and work in overcrowded, unsanitary conditions for extended periods, increasing vulnerability to STH infections such as Strongyloides and hookworm [10–12].

Incorporating variables capturing duration of stay, housing type, and occupational environment could substantially improve the model’s predictive capacity and relevance to real-world migrant populations.

3. Ethical dimensions and risks of stigmatization

While nationality-based risk profiling may guide efficient screening, it raises ethical concerns. Such profiling can unintentionally reinforce stereotypes or exacerbate xenophobia [13]. Current WHO and UNHCR guidance emphasize voluntary, informed screening with culturally sensitive communication to avoid discrimination and exclusion, especially for undocumented migrants [14].

Therefore, predictive tools should be implemented alongside robust community engagement and safeguards to uphold equity and human rights.

4. Generalizability and contextual calibration

The diversity of migrant populations across regions necessitates model adaptation. In northeastern Thailand, we found STH risk correlated more strongly with overcrowding and gaps in deworming coverage than with national socioeconomic indicators [15]. Moreover, migrant demographics differ substantially between Europe and Southeast Asia—for example, migrants in Italy tend to be recent arrivals and predominantly male, whereas Southeast Asian urban migrants may include long-term residents and second-generation individuals [16].

Tailoring models to local epidemiological and demographic contexts is essential to optimize their utility and accuracy.