A Copenhagen-area municipality piloted a points system assigning up to 3,000 risk points for parental mental illness, or 300 for a missed dental visit, to flag children for early intervention. Denmark's DPA blocked it in 2019; an ombudsman inquiry into it opened in 2024.
3,000
Risk points assigned for parental mental illness (circa 2018)
1,000
Risk points assigned for a missed medical appointment (circa 2018)
500
Risk points assigned for unemployment (circa 2018)
300
Risk points assigned for a missed dental appointment (circa 2018)
3 (Gladsaxe plus two unnamed partner municipalities)
Municipalities piloting the model (circa 2018)
Details
Maturity
Discontinued
Promoter
Gladsaxe Kommune (Municipality), with two unnamed partner municipalities
Period
Piloted circa 2018; work halted 2019 after Datatilsynet denied permission; subject to a Parliamentary Ombudsman inquiry opened June 2024
Region (NUTS)
DK01
Keywords
child protection, social services, predictive risk scoring, welfare data ethics
Context
Around 2018, the municipality of Gladsaxe, on the northern outskirts of Copenhagen, piloted a predictive risk-scoring tool — since known as 'the Gladsaxe Model' — alongside two other unnamed Danish municipalities.
Objectives
The tool assigned numeric points to family circumstances to flag children for early social-services intervention: parental mental illness scored 3,000 points, unemployment 500, a missed medical appointment 1,000, a missed dental appointment 300, and parental divorce also contributed to a family's cumulative score. Families crossing a threshold were to be referred for closer follow-up by child-welfare caseworkers.
Activities
A separate follow-on research project at Aarhus University that attempted a similar approach drew comparable criticism.
Results
Denmark's data protection authority, Datatilsynet, denied the municipalities permission to continue processing data this way in 2019, and following critical press coverage the pilot was halted without the municipality publishing any outcome or accuracy statistics — the complete absence of published performance data is itself one of the model's most consistent independent criticisms.
Conclusions
The Gladsaxe Model resurfaced in November 2024 when Amnesty International's 'Coded Injustice' report on Denmark's automated welfare state cited it as a named example of opaque risk-profiling in child protection, warning it and similar tools risked discriminating against disabled, low-income, migrant and minority families. Denmark's Parliamentary Ombudsman opened a formal inquiry into the Gladsaxe Model and related systems in June 2024, examining discrimination concerns and citizens' notification rights; that inquiry remained open as of this writing.
Implementation
Indicative cost
Low (< €50k) — No budget or cost figures published for the pilot.
Time to results
Short (< 1 year) — Piloted circa 2018; halted 2019 after Datatilsynet denied permission to continue; Parliamentary Ombudsman inquiry opened June 2024, remained open as of this writing.
Staffing & skills
Gladsaxe Kommune child-welfare caseworkers, two unnamed partner municipalities
Conditions for success
publishing accuracy/outcome statistics
transparent notification of families being scored
data protection authority approval before processing
Common failure modes
Denmark's data protection authority (Datatilsynet) denied permission to continue processing data in 2019
no outcome or accuracy statistics were ever published — a gap independent watchdogs (AlgorithmWatch) specifically flag as a core criticism
Amnesty International's 2024 'Coded Injustice' report cited it as opaque risk-profiling risking discrimination against disabled, low-income, migrant and minority families
a follow-on research attempt at Aarhus University drew comparable criticism
Denmark's Parliamentary Ombudsman opened a formal inquiry in June 2024, still open, over discrimination concerns and citizens' notification rights
Where it fits
Governance type
municipal government, child welfare services
Scale
3 municipalities piloted, Denmark
Income level
high-income (Denmark)
Commonly funded by
National / regional programmes
Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.
Do you run this practice?
Claim it —
verified implementers get a public contact pathway and can propose corrections.
Data sources
Where this practice's information was retrieved from, and when.
Since 2019, Chile's Subsecretariat of Social Evaluation has piloted a machine-learning risk score prioritising at-risk children in 12 communes. Seven …