Informal care assistant

Municipality of Assen

The Mantelzorg Assistent offers personalised support to informal carers. The algorithm collects and analyses data entered by informal carers, such as demographics and care severity, to provide targeted advice and information. The aim is to ease the burden of informal care and provide policy insight to the municipality.

Last change on 24th of April 2025, at 7:34 (CET) | Publication Standard 1.0

• Publication category: Impactful algorithms
• Impact assessment: DPIA, IAMA
• Status: In use

General information

Theme

Health and Healthcare

Begin date

2024-05

Contact information

[e-mail]

Link to publication website

https://assen.nl/privacyverklaringavg-algemene-verordening-gegevensbescherming

Responsible use

Goal and impact

Aim: The aim of the Mantelzorg Assistent is to provide customised information and access to care for informal carers. It also helps the municipality of Assen to gain better policy insight into the informal care population. The ultimate goal is to ease the burden of informal care, improve communication and coordination, increase the knowledge and skills of informal caregivers and promote their well-being.

Impact: The algorithm has a significant impact on supporting informal carers. By providing personalised information and assistance, it helps informal carers to be better informed and supported in their care task. This leads to improved quality of care and increased well-being of both informal carers and those in need of care. In addition, the algorithm provides the municipality with valuable insights into the needs and characteristics of the informal caregiver population, which helps formulate effective policies and targeted support.

Considerations

Benefits:

The algorithm provides carers with personalised information and support, helping them more quickly and appropriately. This improves the quality of care and increases the well-being of both informal carers and those in need of care. In addition, the algorithm allows for more efficient communication and coordination, saving time and resources. The data collected gives Assen municipality valuable insight into the needs of informal carers, allowing for more targeted policy development. Informal carers experience more peace of mind thanks to targeted advice, making them feel better supported.

Disadvantages:

There is a risk of bias, potentially favouring or disadvantaging certain groups of informal carers. Processing personal data also poses privacy risks, such as unauthorised access to sensitive information. Furthermore, informal carers with limited digital skills may have difficulty using the app, which could lead to unequal access to support.

Alternatives:

Manual support by municipal staff is possible, but less efficient and less personalised. General provision of information via, for example, brochures or websites does not offer customisation and is less responsive to the individual needs of informal carers.

Ethical considerations:

Ensuring privacy and data protection is essential; this is addressed through a DPIA and compliance with the AVG, with appropriate security measures. To ensure equal treatment, the algorithm is regularly checked for bias and audits are carried out. Transparency is provided by clearly informing carers about the use of the algorithm and the processing of their data.

Conclusion:

The algorithm offers significant benefits for the support of informal carers and the policy insight of Assen municipality. Although there are risks, these are mitigated by appropriate privacy, equal treatment and transparency measures. The advantages outweigh the disadvantages, justifying the use of the algorithm and contributing to the well-being of informal carers.

Human intervention

Use of algorithm outcomes:

The results of the algorithm are used by staff of the Assen municipality and other organisations involved, such as Vaart Welzijn. These staff use the information to better support informal carers and make policy.

Monitoring and adjustment:

1. Human control: Employees check the results of the algorithm by comparing the data with their own knowledge and experience. They can also seek feedback from informal carers to see if the advice and information are useful.
2. Adjustment: If it turns out that the algorithm does not work well or gives wrong advice, the staff can adjust the algorithm. This is done by adjusting the data or re-training the algorithm with new information.
3. Regular audits: Regular audits are carried out to check that the algorithm is still working properly and not making mistakes. This helps detect and resolve any problems quickly.

No human intervention:

In some cases, no human intervention is needed. For example, if the algorithm automatically sends messages to carers with tips and advice, this is done without the need for an employee to check it.

Risk management

Technical risks:

The data security of carers is ensured by encryption, secure storage and the use of SSL/HTTPS for secure communication. The supplier is responsible for maintaining systems and software, with regular updates and checks to ensure the security and functioning of the algorithm. The functioning of the algorithm is continuously monitored so that any errors are quickly detected and resolved.

Legal risks:

The algorithm complies with the AVG and a DPIA has been carried out to identify and mitigate privacy risks. Carers are clearly informed about the use of their data and the operation of the algorithm; an algorithm statement is available. In addition, informal carers explicitly consent to the use of their data within the app.

Financial risks:

The development and maintenance costs of the algorithm are reviewed regularly to ensure efficient use of resources. There is a set budget for the project, with periodic reviews to ensure spending is within limits.

Ethical risks:

To avoid discrimination, the algorithm is regularly checked for bias through audits and monitoring. Transparency and explainability are key principles: carers receive clear information on how the algorithm works and how their data is used. Efforts are made to treat all informal carers fairly and objectively, regardless of background or digital skills.

Periodic monitoring:

Regular audits ensure the correct functioning of the algorithm and help resolve any issues quickly. The impact and performance of the algorithm are evaluated periodically, followed by improvements where necessary. Feedback from carers and employees is actively collected and used to further optimise the algorithm.

Legal basis

The legal basis for the process in which the algorithm is deployed is the Social Support Act (Wmo). This act aims to allow people to live independently at home and participate in society for as long as possible. The Wmo regulates that municipalities are responsible for providing support to people who need help, such as the elderly, people with disabilities, and informal carers.

Purpose of the Wmo:

1. Independence: The law helps people live and function independently in their own environment for as long as possible.
2. Participation: The law encourages people to actively participate in society, for example through volunteering or social activities.
3. Support: Municipalities must ensure that people who need help, such as informal carers, get the right support. This can range from domestic help to counselling and day care.

The algorithm is used to better support informal carers and to gain policy insight into the informal carer population within the municipality of Assen. By using the algorithm, the municipality can provide more efficient and targeted support, which is in line with the goals of the Wmo.

Links to legal bases

Wet maatschappelijke ondersteuning 2015: https://wetten.overheid.nl/jci1.3:c:BWBR0035362&z=2025-01-01&g=2025-01-01

Link to Processing Index

niet gepubliceerd

Impact assessment

• Data Protection Impact Assessment (DPIA): niet gepubliceerd
• Impact Assessment Mensenrechten en Algoritmes (IAMA): niet gepubliceerd

Operations

Data

Personal details:

1. Mail address: Entered by informal carers themselves.
2. Username: Entered by informal carers themselves.
3. Age category: Optional, entered by informal carers themselves.
4. City of residence: Optional, entered by informal carers themselves.
5. Family composition: Optional, entered by informal carers themselves.
6. Severity of care: Optional, entered by informal carers themselves.
7. Occupation: Optional, entered by informal carers themselves.
8. Health data: Optional, entered by informal carers themselves.

Sources of data:

1. Informal carers themselves: Data are entered directly by informal carers via the app.
2. Municipality of Assen: General information and policy data.
3. Partners such as Alzheimer Nederland, Dokter Drenthe, Epilepsie Nederland: Specific information and content for informal carers.
4. Insurance companies, VEGRO, Medipoint: Information on care products and services.

Use of data:

1. Algorithm training: The data entered is used to train and improve the algorithm.
2. Personalised support: The algorithm analyses the data to offer targeted advice and information to carers.
3. Policy insight: The data collected helps Assen municipality to better understand the needs of informal carers and formulate targeted policies.

Technical design

Input: The algorithm uses different types of data entered by informal carers themselves through the app. These data include:

• Mail address
• Username
• Age category
• City of residence
• Family composition
• Care category
• Profession
• Health data

In addition, the algorithm uses information and content from partners such as Alzheimer Nederland, Dokter Drenthe, Epilepsie Nederland, insurance companies, VEGRO, Medipoint, and general policy data from the municipality of Assen.

Functioning: The algorithm is a self-learning model that uses machine learning techniques, including neural networks and collaborative filtering. The algorithm goes through the following steps:

1. Data collection: The entered data from informal carers is collected and stored in a secure database.
2. Data analysis: The algorithm analyses the data to identify patterns and trends. This uses techniques such as clustering and classification to categorise the data.
3. Training: The algorithm is trained using a training set of data. Here, neural networks are used to optimise the model and improve accuracy.
4. Validation: The model is validated using an independent test set to assess its performance and accuracy. This involves looking at metrics such as precision, recall, and F1 score.
5. Adjustment: Based on the validation results, the model is adjusted and optimised. This may mean adjusting certain parameters or re-training the model with new data.

Output: The output of the algorithm consists of personalised advice and information for informal carers. This includes:

• Tips and advice based on the carer's specific situation and needs.
• Periodic messages with relevant information, such as reminders to take in enough fluids in hot weather.
• Policy insights for the municipality of Assen, such as statistics and trends on the informal care population.

External provider

Valtes Care B.V.