A while ago, in preparation of the upcoming review by the UN Committee on the Rights of Persons with Disabilities (UN CRPD Committee) on the Netherlands, I have been looking into some worrying dynamics in Dutch mental health care. Some points deserve to be highlighted.
I would like to share my brief reflection on data collection, algorithms, profiling, and risk assessment in Dutch mental health care. At present, developments regarding machine learning and Artificial Intelligence (AI) are increasing the risk of abuse of data to perpetuate discrimination.
Data collection and statistics
Since the Dutch child care benefits scandal[1] in 2021, the Netherlands is by now known for its wrong use of statistics.
Statistics can only be used for data collection, and they do not have a predictive value regarding people or their behavior (that would be stereotyping, profiling, amounting to arbitrariness and discrimination. If 99 persons choose to do a certain thing, it still does not mean that the next person would do the same).
“Risk assessments” (in Dutch: “risico taxatie”)
Risk assessments are used in mental health care world-wide, also in institutions in the Netherlands[2], and especially in forensic services[3]. There is a huge variety of available risk assessment tools and their way of assessing[4]. Clinical risk assessment tools (e.g. HKT, HCR20V3, FARE) are based on weighing the risk factors by the assessor, while actuarial risk assessment are based on algorithms (chance calculation). In the Netherlands, the common understanding seems to be that actuarial risk assessment is suitable for ‘group risk’, while clinical risk assessment is seen as suitable for individual risk. In the Netherlands, clinical risk assessment is most used, e.g under the claim of being able to include individual parameters and ‘low prevalent risk factors’.
Risk assessments are often meant to identify e.g. the risk of violence, abuse, recidivism, or suicide risk. Clinical risk assessments often use static factors e.g. age, gender, background, history which can influence the outcome, yet these cannot be changed by the person, so they score ‘points on the risk scale’ for being who they are).
Profiling
Before World War 2, the psychiatric profession also claimed to be able to predict criminality and personal traits (through eugenic theories such as craniometrics, phrenology and anthropological criminology, and ‘scientific racism’).
Now in the 21st century we have “risk assessments” that rob people of their freedoms based on algorithms, stereotyping, and ‘personal weighing of a subjective assessment’. A person with a history of child abuse will always score higher in a clinical risk assessment.
The risk assessments can cause forensic sentences to last without a fixed end date, because re-evaluation leads to the same results.
Ableism
The mental health sector in the Netherlands and beyond is in fact characterized by a strong focus on ‘normal’ and ‘abnormal’ behavior, and over the past centuries up to now, mental health care generally aimed to ‘delete abnormal’, which is actually showing ableism (validism). Rejecting ‘abnormal’ and considering it as a “risk”, is not the same as providing support for inclusion.
Presumed dangerousness linked to psychosocial disability
In an evidence based justice system, deprivation of liberty cannot depend on subjective decisions and forecasts of the future such as “presumed dangerousness”.
The widespread claims that persons with psychosocial disabilities may bear a supposed “risk of posing a danger to self or others, causally linked to psychosocial disability” are clear forms of discrimination, and this structure is legally and morally baseless for many reasons.
Besides the fact that “risk” and “danger” are no objective values, nonetheless they invoke negative associations (e.g. fear). These stereotypes cannot be used to justify deprivation of liberty or other rights. Such systems would amount to e.g. arbitrary detention and other serious violations of universal human rights. Deprivation of rights has got nothing to do with providing support. Deprivation of rights on the basis of psychosocial disability is a violation of international human rights treaties, such as articles 12 and 14 of the UN Convention on the Rights of Persons with Disabilities. Therefore, the Dutch laws WVGGZ, WZD and WFZ need to be withdrawn.
Furthermore, the suggestion of a “causal link” between supposed dangerousness and psychosocial health status promotes very stigmatizing negative stereotyping. This supposed causal link clearly does not correspond with the actual facts and statistics (e.g. many persons with psychosocial disabilities do not ‘pose a danger’). There is actually no “causal link” between supposed ‘dangerousness’ and psychosocial disability.
The entire sequence of legislation and practices that follow the outdated concept of presumed dangerousness are founded in discrimination, and are in violation of the UN Convention on the Rights of Persons with Disabilities.
Overall, the concept of presumed dangerousness based on psychosocial disability is thriving on negative stereotypes, and has given rise to mental health laws that enable a large series of human rights violations worldwide, and as such, the suggestion or presumption of ‘danger’ linked to psychosocial disability should be considered as hate speech, which can cost lives.
Equality and non-discrimination
As is the case for any person, innocence must be presumed.
Deprivation of rights on the basis of assumptions is a violation of international human rights law.
In addition, the concept of “a mental disorder of such a severity that it warrants treatment, even against the persons will” is an outdated ableist concept and in violation of human rights standards. The principle of respect to diversity implies that psychosocial diversity should never be a reason for any deprivation of rights. Persons with psychosocial disabilities are entitled to enjoy all human rights on a full and equal basis with others, including by exercising legal capacity, liberty, being free from torture, violence, abuse and exploitation, and exercising their right to free and informed consent to health care.
‘Ableist pseudo-science’
Risk assessment tools are not a solution to ‘reduce subjectivity’ in decision making. Risk assessments are based on stereotypes (by using statistics to take rights away), combined with ableism/validism (e.g. enforcing negative stereotypes), and therefor these tools are inherently discriminatory. These tools are not oriented on fostering health, wellbeing or recovery, nor on the protection of human rights. These tools and concepts are focused on negative stereotypes, serving to impose exclusion and restrictions. They can result in permanent confinement in institutions.
And since risk assessment tools are based on pervasive “assumptions of danger”, they are inherently in violation of the purpose and the universal cornerstones of the human rights principles, such as equality and non-discrimination, respect for inherent dignity and individual autonomy including the freedom to make one’s own choices, and independence of persons, and full and effective participation and inclusion in society.
The risk assessment tools and baseless presumptions of ‘danger caused by mental illness’ are in fact part of ‘ableist pseudo-science’ which is a form of biased pseudoscience within a tunnel vision of ableism.
The “risk paradigm” including its risk assessment tools in mental health care should be illegal, since they violate human rights and contribute to stigma and discrimination.
Inclusive development
Perhaps needless to say, yet, there are several reasons to be concerned about the development of machine learning and AI,which should not perpetuate the same ableist stereotyping and discrimination.
The ableist concepts on risks and dangers associated with persons with psychosocial disability have been a barrier for the realization of equal rights of persons with psychosocial disabilities for ages.
The UN CRPD brings a long awaited clarity on the universality of human rights, and underlines that persons with disabilities are entitled to enjoy the same rights as others. Any discrimination on the basis of disability is prohibited. This means that forced treatments on persons with psychosocial disabilities are a violation of international law, and that the “risk paradigm” needs to end, and be replaced with a “support paradigm”.
What is needed is a focus on support for inclusion, instead of exclusion.
And instead of a risk-assessment, the focus should be on supporting individual recovery.
Real contact is more valuable than any assessment.
The big question regarding mental health care is now:
The use of algorithms for ‘risk assessment’ among the population has shown to be discriminatory in the child care benefits scandal in the Netherlands (which caused resignation of the Dutch cabinet in 2021).[5]
- Will the use of risk assessment tools on the prediction of behavior of persons in mental health care, including forensic psychiatry, now be reconsidered?
Risk lists: Top 600 / Top 400/ Top 1000
Also at the community level, in a number of Dutch cities there are designated tranches to deal with young persons who have been in contact with the police for , to offer them a so called combination of care and punishment. To that end, these children and youngsters were placed on a list, the local “Top 600 list” which implies that these kids are regarded as likely to end up in criminal affairs. In several places the “Top 600 list” was quickly filled up, and in several places the list was expanded up to “Top 1000” and/or adding a “Top 400” which targets additional youngsters who may be “the next Top 600”. While Top 600 has a certain threshold of criminal activity, Top400 is focusing on much lighter situations, e.g. for a minor offense, such as theft of a can of coke, or for being close with a person who is listed in Top-600.
These lists obviously work with “risk assessments”, and they assess kids and youngsters as “High Risk” based on certain characteristics (e.g. life circumstances, parental circumstances, and ethnicity seems to be a factor too) (Top 400/ Top 1000/ Top 600). Parents report on various media[6] that they were suddenly informed that their child was on a list, and many people report of negative consequences[7] of being targeted through such a list. (e.g. organizations suddenly act different towards them, and these kids are branded).
Also children in Amsterdam who have not done anything wrong can be targeted by the Top 400 risk list[8]. That is partially due to the fact that the Municipality in Amsterdam has developed an algorithm which predicts the future behavior of youngsters. The algorithm called “ProKid” targets young people before they have committed any crime. They can be on the list because of the actions of someone else, or because they changed of school several times. Yet, these targeted kids were then de facto treated as young offenders. This is stigmatizing the kids who may actually need support.
In Amsterdam, the use of the algorithm “ProKid” and other algorithms has stopped, yet, the government wants a strict policy against crime, and they maintain very broad criteria which may easily consider anyone as posing a ‘risk’.
The risk of risk assessment
Risk assessments in its many forms have already ruined many persons and many families.
The “ProKid algorithm” seems to be the same type of “risk assessment” based on stereotypes and profiling, and again maintaining a similar kind of assumptions on supposed ‘dangerousness’ of certain groups. A similar use of algorithms was already found discriminatory in the Dutch child care tax scandal in 2021.
The discriminatory risk assessments are in use at many places.
Its use is spreading and must be stopped.
*
Risk assessments which focus on predicting people’s future behavior need to be abolished fully, since they always include profiling and stereotyping, and are acts of discrimination.
(and presumably only non-living objects or dynamics, such as goods and processes, can be subject to algorithms without violating human rights).
[1] https://www.amnesty.org/en/latest/news/2021/10/xenophobic-machines-dutch-child-benefit-scandal/
[2] 2013 https://www.researchgate.net/publication/257628861_Het_gebruik_van_risicotaxatie-instrumenten_onder_SPV-en_The_use_of_risk_assessment_tools_by_Dutch_social_psychiatric_nurses
[3] https://www.forensischezorg.nl/beleid/risicotaxatie-instrumenten
[4] https://www.tijdschriftvoorpsychiatrie.nl/assets/articles/60-2018-8-artikel-vanderput.pdf
[5] https://www.amnesty.org/en/latest/news/2021/10/xenophobic-machines-dutch-child-benefit-scandal/
[6] https://www.tienersachtertralies.nl/hoofdstuk-10
[7] https://www.trouw.nl/binnenland/het-verhaal-van-top-400-kinderen-hij-stal-alleen-een-colablikje~b7f42772/
[8] https://www.trouw.nl/binnenland/ook-jongeren-die-niets-hebben-uitgehaald-komen-op-amsterdamse-risicolijst-top-400~b4d1680d/
Welkom op de weblog van Actiegroep Tekeer tegen de isoleer! 