PhD: Using Data and AI to Improve Child Health through Housing

PhD: Using Data and AI to Improve Child Health through Housing

Housing features may predict child-health risks, but which housing changes would actually improve outcomes? This TU/e doctorate connects quantitative urban research, statistical/AI modelling and medical collaboration to investigate pregnancy and early-childhood housing conditions. A scenario-based tool is an intended research output, not a validated clinical product.

The team has relevant retrofit-evaluation and housing-data experience. The decisive project-specific question is which child, health and housing data will be available and authorised for use. The employer advertises four years with a nine-month assessment, EUR 3,204–4,051 gross monthly full-time base salary and 0.8–1.0 FTE. How part-time work affects calendar duration and workload remains to be confirmed.

The employer vacancy notice lists 15 November 2026 as the last application date and says the position remains open until filled; current application status still requires confirmation. Prepare the online cover letter and CV with publication list and three referee contacts, while asking about data access, causal scope and academic registration. The advert states no exact closing clock or cover-letter page limit. Official employer vacancy 2026/644

Research Problem and Doctoral Work

From housing records to a defensible health question

Poor insulation, heating and ventilation may affect health, but a model that predicts an outcome from housing features does not automatically identify the benefit of changing those features. Household income, prior health, neighbourhood and the selection/timing of renovations can all complicate interpretation. The vacancy explicitly targets pregnancy and early childhood, uses medical and behavioural insights, large-scale data, computer vision and advanced statistics, and aims to feed predictions into a scenario-based digital tool for housing providers and municipalities. These are research goals; the tool and its health benefits are not established results. TU/e official vacancy 2026/644: Data and AI for child health through housing

The advertised programme is Homes for Upcoming Generations, described by the employer as NWO-funded. Scientific partners are Erasmus Medical Centre, Erasmus University Rotterdam and VU University Amsterdam; social partners include housing providers, municipalities and engineering bureaus. No dataset names, sample sizes, image sources, linkage keys, collection schedule, trial protocol, grant number or access approvals are specified. Cooperation with a medical centre does not establish that the candidate will recruit children, deliver treatments or have unrestricted clinical records. TU/e official vacancy 2026/644: Data and AI for child health through housing

Methods and possible contributions

The following are discussion proposals, not assigned work packages. One line could validate extraction of housing conditions from images or documents against human-labelled records, with error analysis across building types. Another could separate predictive risk from intervention effects, specifying exposure windows during pregnancy/early life and testing assumptions behind renovation comparisons. A third could evaluate scenario decisions under uncertainty, cost and equity constraints. Validation should hold out time periods or housing providers, check calibration and missingness, and avoid leaking records of the same family or dwelling across model splits. The actual causal design and computable outcomes depend on available data. TU/e official vacancy 2026/644: Data and AI for child health through housing; Health-evaluation precedent; Unstructured-data precedent; Behavioural/econometric precedent

Data governance is part of feasibility. Prior group research used Statistics Netherlands microdata, but this is not proof that the new project has the same access. CBS requires institutional authorisation and project review, confidentiality commitments and researcher training; results undergo disclosure review before release. If CBS data are used here, establish approved datasets, researcher access, permitted software and model/output export. Likewise, an earlier social-media study's ethics approval does not cover a new project involving children or health records. Energy-efficient homes: effects on poverty environment and comfort; Statistics Netherlands microdata research access procedure; Does online sustainability communication shape public discourse? Insights from six years of tenant-housing provider interactions

School, Programme and Field Context

The employing institution is Eindhoven University of Technology (TU/e), Netherlands, and the vacancy is in the Department of Built Environment, Urban Systems and Real Estate group. The institutional research portal uses Real Estate and Urban Development for Ossokina's unit; that naming difference should not be silently turned into a confirmed organisational restructuring. This is an advertised TU/e doctoral position. Erasmus MC/EUR/VU have collaboration roles; there is no evidence of a joint/double degree or separate employment with these partners. TU/e official vacancy 2026/644: Data and AI for child health through housing; Ioulia V. Ossokina official research profile; TU/e Real Estate and Urban Development institutional unit

The four-year arrangement includes research training, seminars and international conferences, plus teaching: at least 10% of the four-year employment overall, with a maximum 15% per year. These are distinct limits, not a fixed 10% each year. The employer's four-year full-time wording coexists with the 0.8–1.0 FTE field; confirm whether 0.8 FTE changes the calendar duration, workload and completion plan. Formal academic admission, registration, training credits, dissertation requirements and any employee-specific fees should be confirmed with TU/e doctoral services; a job offer should not be treated as all academic procedures completed. National Nuffic guidance distinguishes employed doctoral researchers from grant/external routes, and only research universities award Dutch PhDs. Its general fee discussion is not TU/e's individual fee decision. TU/e official vacancy 2026/644: Data and AI for child health through housing; Study in NL doctoral degree funding and admission framework

For institutional context, the cited ShanghaiRanking profile lists ARWU 2026 overall 301–400, GRAS 2026 Civil Engineering 46, and Artificial Intelligence 101–150. The latter two are subject-wide indicators relevant to built-environment/data work, not measures of this team's public-health expertise, mentoring or doctoral completion. The more direct research evidence lies in the papers below and the university's interdisciplinary urban-economics line. ShanghaiRanking TU/e 2026 institutional and subject profile; Ioulia V. Ossokina official research profile

Supervisor and Lab Assessment

Roles and research environment

Ioulia V. Ossokina is the named academic contact and an Associate Professor in Built Environment, with EIRES and EAISI Health affiliations. Her background is mathematical economics/econometrics, including a 2003 Erasmus PhD and prior CPB economic-policy research. Her programme combines mathematical models, big data, AI and behavioural methods to analyse housing, poverty, health and comfort, translating findings into decision tools. This is a relevant intellectual home; the vacancy does not yet identify the formal promotor, co-supervisor allocation, daily advisor or meeting frequency. Ioulia V. Ossokina official research profile

The vacancy calls Leo van Houten a hiring manager, while his official personnel page lists an HR advisor role in Chemical Engineering and Chemistry. He should not be presented as an academic supervisor, and that HR affiliation does not override the post's Built Environment location. Prior BEL work establishes experience with housing providers and empirical/structural modelling. Its NWO grant 403.19.230 and named partners are historical research context, not the funding number or approved data package for Homes for Upcoming Generations. TU/e official vacancy 2026/644: Data and AI for child health through housing; Leo van Houten official staff role; BEL Behaviour Energy transition Low income project

Recent papers: contributions and limits

Roberdel, Ossokina, van Ommeren and Arentze, The Lancet Public Health, 2026. The structured abstract reports a Dutch retrofit natural experiment following two million people over ten years, comparing 180,000 retrofitted tenants with not-yet-retrofitted controls using staggered difference-in-differences and individual fixed effects. Among under-18s, respiratory medication use fell by 3.76% relative (95% CI 1.04–6.48; p=0.0067). The reported five-year asthma-medication reduction of 6.91% has a confidence interval crossing zero (−0.04–13.85; p=0.051), so it should not be described as conventionally statistically significant. Non-respiratory medication and healthcare-cost effects were not significant. These are bounded medication-use findings, not proof of a new AI tool, pregnancy effects or a clinical cure. The available evidence is the institution's structured abstract; detailed full-text methods and appendices remain unconfirmed. Effect of energy-efficient homes on residents health: evidence from a natural experiment in the Netherlands

Van Casteren, Ossokina and Arentze, Energy Research & Social Science, 2024. A panel-logit analysis of 78 Dutch community retrofit campaigns and roughly 66,000 owner-occupied dwellings studies block leaders and adoption. Associations vary with proximity, technology and other campaign activity; the strongest neighbourhood effects are within 200 metres. This supports behavioural questions about whether proposed upgrades are adopted. It is not a child-health experiment, and the authors leave the exact propagation mechanisms open. Confidential source data are not automatically available to a new doctoral researcher. The evidence cited here covers selected data, model, results and conclusion sections of the full article. Do you listen to your neighbour? The role of block leaders in community-led energy retrofits

Juneja, Verberne and Ossokina, arXiv preprint, July 2026. The study analyses 792 posts and 3,197 comments from 92 housing providers. It combines transformer-based intent/sentiment classification, clustering and multinomial logistic regression. The methods distinguish 246 development and 262 held-out test comments; reported test accuracy is 0.85 for intent and 0.72 for sentiment. This is a concrete example of turning unstructured housing-related text into analysable data, with human labels and synthetic-data augmentation. The authors explicitly describe an observational, single-platform study: it cannot prove communication causes outcomes, nor validate a computer-vision or child-health model. The available HTML full text covers data and ethics, selected methods, results and limitations; the work remains a preprint. Does online sustainability communication shape public discourse? Insights from six years of tenant-housing provider interactions

A further 2023 Tinbergen working paper by Roberdel, Ossokina, Karamychev and Arentze follows 125,000 households, combining quasi-experimental retrofit comparisons with a calibrated consumer-choice model. Average gas savings of 22% versus 16% among the poorest illustrate why comfort, energy and welfare objectives can diverge. It motivates heterogeneous responses and behavioural modelling; it does not measure the new project's child-health effect. The evidence cited here focuses on the paper's identification, model and conclusion sections. Supervision-completion and graduate-placement outcomes remain unconfirmed. Energy-efficient homes: effects on poverty environment and comfort

Funding, Application and Living Cost

The employer advertises a four-year paid position with a nine-month intermediate assessment. Full-time scale-P base salary is EUR 3,204–4,051 gross per month, with 8% holiday allowance and 8.3% year-end bonus based on annual gross base pay. The range should not be treated as the first-year salary choice or take-home income. The 2026–2027 university agreement provides the collective framework; do not add its pay increase again to the vacancy's stated rates or assume a new hire qualifies for past one-off payments. TU/e official vacancy 2026/644: Data and AI for child health through housing; Final Dutch Universities collective agreement 2026–2027

The vacancy states 29 leave days on a 38-hour week, expandable to 41 through a 40-hour arrangement, prorated for part-time work. ABP pension premiums are split 70% employer/30% employee; the latter is a share of the premium, not 30% of gross salary. Advertised benefits include parental/maternity provisions, training, commuting/home/internet allowances, wellbeing support, affordable sports and campus childcare. They do not guarantee a childcare place or personal tax compensation. Exact start date, pay step, 0.8 FTE contract length, doctoral charges, relocation/data/compute/travel budgets and support if completion overruns require clarification. TU/e official vacancy 2026/644: Data and AI for child health through housing

Nuffic's national student reference is EUR 1,000–1,500 monthly, citing 2024 itemised data; it is not an employed researcher's Eindhoven budget. The regional governmental Expat Center's recorded indicative private rents, excluding utilities, are EUR 1,100–1,400 for a furnished studio and EUR 1,400–1,750 for a furnished one-bedroom apartment. A private apartment can therefore consume a large share of net pay. Shared accommodation may differ, but no low-rent unit is reserved in the job. Budget actual rent, utilities, health insurance, tax/pension deductions, deposits and arrival costs; no personal savings estimate is supported. Study in NL cost of living reference; Holland Expat Center South: renting and Eindhoven private rents

The employer vacancy notice lists 15 November 2026 as the last application date and says it remains open until filled. Use the employer's 15 November deadline and submit ahead of the last day. No exact clock time is confirmed on the employer page. Applications are online only, comprising a cover letter and CV including publications and contact details of three references. References may be approached at any stage; the employer recommends notifying them. The advert gives no cover-letter page limit. TU/e official vacancy 2026/644: Data and AI for child health through housing

Applicant Fit and Preparation

The required MSc fields are exact/data sciences, quantitative economics or engineering. A strong record in several of calculus, linear algebra, computational modelling, microeconomics, econometrics, operations research and data science, demonstrated Python and big-data experience, analytical skills, spoken/written English and interest in industry collaboration are expected. Working experience at a European university/research institute is an advantage, not a European-nationality requirement. The vacancy requires neither a prior doctorate nor a medical degree, Dutch-language certificate or clinical licence. It specifies no IELTS/TOEFL cutoff or exemption; obtain TU/e's applicable documentary requirement rather than borrowing another university's threshold. TU/e official vacancy 2026/644: Data and AI for child health through housing

A useful portfolio would show a reproducible quantitative project: data provenance/cleaning, missingness, validation and interpretation, with a clear distinction between prediction and causal inference. For a housing-health proposal, identify the population, exposure window, comparison strategy and outcome; explain what cannot be inferred from image/text features alone. These are analytical preparation suggestions, not additional official admission conditions.

International feasibility has positive support: this exact vacancy offers a Staff Immigration Team and international tax arrangements, while IND's non-EU researcher route explicitly includes paid researchers and doctoral candidates. A recognised sponsor, appropriate higher-education qualification, hosting/employment documents and other conditions still apply. This supports in-principle consideration of Chinese applicants, subject to individual degree, language, immigration and employer checks. The advertisement also permits pre-employment screening, including knowledge security. Its linked 2022 guidelines advocate proportionate risk assessment and avoiding arbitrary exclusion; they are not individual clearance or proof of the law's 2026 status. The vacancy does not state a categorical ban based on Chinese nationality. TU/e official vacancy 2026/644: Data and AI for child health through housing; IND researcher residence permit Directive (EU) 2016/801; National Knowledge Security Guidelines 2022 linked by employer; Study in NL doctoral degree funding and admission framework

Because the role is advertised in quantitative/data disciplines, clinical practice should not be invented. IND's BIG-registration condition applies if someone actually practises a regulated healthcare profession. Ask whether this project's participant contact, Dutch records and stakeholder meetings require Dutch support, and whether any clinical duties are envisaged. The absence of a Dutch requirement in the advert does not establish that every dataset, meeting or teaching task will be English. TU/e official vacancy 2026/644: Data and AI for child health through housing; IND researcher residence permit Directive (EU) 2016/801

Field and Career Context

Housing renovation sits at the intersection of building analytics, public policy, population-health research and decision support. The European Commission's Renovation Wave prioritises renovation, energy poverty, digital approaches and technical assistance for local actors. This is evidence of a sustained policy/application setting, not current job vacancies or confirmed financing for this doctorate. Possible paths include urban/building data research, policy evaluation, housing-provider analytics and academic research; this evidence does not quantify individual placement or salary. European Commission Renovation Wave

Next Actions and Questions to Verify

Before applying: submit the complete online file before 15 November; prepare three referee contacts and evidence of Python/quantitative work. Ask Ossokina which datasets and approvals are ready, which are planned, and whether the research scope can proceed if a linkage is delayed. Confirm foreign-Master recognition and English documentation with doctoral/HR services.

At interview: establish the formal promotor/co-supervisors, daily guidance and the division between computer vision, statistics, causal evaluation and tool development. Specify pregnancy/child outcome definitions, data-access/ethics responsibilities, validation and publication/model-export constraints. Confirm Dutch-language support and the teaching plan.

Before accepting: obtain the FTE, calendar duration, nine-month review and continuation terms in writing, alongside academic registration, any charges, immigration/screening steps and project expenses. Check a real housing budget against payroll information. The opening offers an unusually concrete housing-evaluation research lineage; the decisive remaining test is whether the new child-health question has accessible, authorised data and a workable doctoral plan.

Evidence and Sources

The cited vacancy and institutional records do not establish the role's current status or an individual applicant's eligibility. Evidence scope differs by paper: the health natural experiment is supported by a structured abstract, without full methods or appendices; the 2024 retrofit paper by selected data, model, results and conclusion sections; the July 2026 preprint by selected HTML sections on data, ethics, methods, results and limitations; and the 2023 working paper by identification, model and conclusion sections. Abbreviated author names are retained without inferring full bylines or additional roles.

The statistical uncertainty, preprint/working-paper status and difference between pregnancy/child outcomes and earlier adult/household/communication studies remain explicit. Earlier CBS access and ethics approvals do not authorise this project. Salary, rent, fee and immigration/security references remain context for confirmation by the employer/university; no personal approval or current legal assessment is supplied.

Position, people and doctoral setting

• TU/e official vacancy 2026/644: Data and AI for child health through housing
• Ioulia V. Ossokina official research profile
• Leo van Houten official staff role
• TU/e Real Estate and Urban Development institutional unit
• BEL Behaviour Energy transition Low income project
• ShanghaiRanking TU/e 2026 institutional and subject profile
• Study in NL doctoral degree funding and admission framework

Research, data and applications

• Effect of energy-efficient homes on residents health: evidence from a natural experiment in the Netherlands
• Do you listen to your neighbour? The role of block leaders in community-led energy retrofits
• Does online sustainability communication shape public discourse? Insights from six years of tenant-housing provider interactions
• Energy-efficient homes: effects on poverty environment and comfort
• Statistics Netherlands microdata research access procedure
• European Commission Renovation Wave

Salary, costs and international eligibility

• Study in NL cost of living reference
• IND researcher residence permit Directive (EU) 2016/801
• Holland Expat Center South: renting and Eindhoven private rents
• National Knowledge Security Guidelines 2022 linked by employer
• Final Dutch Universities collective agreement 2026–2027