AIRe – Multidisciplinary rehabilitation platform

Intended Benefits of the Product 

The AIRe platform consists of two user interfaces: the interactive AIRe Talk interface for clients and the AIRe Hub interface for professionals. AIRe Talk supports clients in their rehabilitation and self-rehabilitation, while AIRe Hub is used to manage rehabilitation content and train the AI. 

The AIRe platform aims to support clients’ functional capacity, self-rehabilitation, and active participation by enabling self-care to begin before an appointment and continue between appointments in accordance with the provided guidance. The aim is also to reduce unnecessary use of healthcare services in situations where self-rehabilitation is sufficient. 

Research Evidence Related to the Assessed Product 

According to a preliminary report, a before-and-after pilot study conducted in Rwanda assesses the implementation and effects of AI-assisted digital rehabilitation using the AIRe platform in community healthcare. AIRe did not operate as a standalone system but was used alongside the Inclusion App to provide AI-based decision support within a human-supervised workflow. 

Ten community health workers used the AIRe platform as part of the rehabilitation care pathway in two regions of Rwanda. The workflow had been professionally validated and was suitable for use in field conditions. Effectiveness was assessed using the EQ-5D-5L health-related quality of life measure and the Visual Analogue Scale (VAS) at baseline and at one-month follow-up. 

The EQ-5D-5L measures health status across five dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. The VAS ranges from 0 (the worst imaginable health state) to 100 (the best imaginable health state). 

Over the one-month period, a clear improvement was observed in participants’ self-reported health status, particularly in pain and functional capacity. The results suggest that AI may improve the consistency of rehabilitation-related decision-making and expand access to services. [1] 

A second cross-sectional study conducted in Rwanda, for which only an abstract was available at the time of the assessment, evaluated the content validity, safety, and feasibility of the AIRe platform. Ten rehabilitation professionals provided structured assessments of the platform. The mean content relevance score was high (4.3/5), and 85% of the assessments were rated as acceptable. From the professionals’ perspective, the solution was considered safe and usable, and its cultural appropriateness was also rated positively overall (85% “yes” responses). Qualitative feedback focused primarily on improving usability and accessibility. [1] 

In addition to the Finnish Digihoitokokeilu pilot programme, AIRe pilots are ongoing in Rwanda. In an upcoming pilot in Finland, the AIRe platform will be used to support hearing and vision rehabilitation, for example by providing guidance related to assistive devices, support, and services. In addition, two research projects are being prepared in which AIRe will support self-rehabilitation for people with low back pain between physiotherapy appointments and provide support to parents or other close persons of children and young people with neuropsychiatric symptoms in challenging everyday situations. [1] 

Literature reviews 

The systematic review included seven studies evaluating the cost-effectiveness of AI-based interventions, particularly for spinal musculoskeletal disorders (MSDs). In this review, artificial intelligence referred to methods and algorithms that mimic human cognitive functions, such as learning, reasoning, and problem-solving. These methods can be used to analyse large and complex datasets, identify relationships between phenomena, and generate predictions or decisions in an adaptive and often autonomous manner. 

The results suggest that AI-based solutions, such as exercise platforms, imaging screening tools, self-management applications, and algorithm-based care pathways, may be more cost-effective than usual care. However, the effects varied depending on the intervention, perspective, follow-up period, and outcome measures. The evidence was limited by heterogeneity across the studies, short follow-up periods, potential publication bias (favouring positive results), and the fact that the studies focused primarily on low back pain. [2] 

A pilot study conducted in Portugal evaluated the use of an AI-based telerehabilitation solution among postoperative patients and patients with chronic musculoskeletal disorders (MSDs) in a university hospital and primary healthcare settings. All patients underwent a clinical assessment, received a treatment plan, and were followed up by a specialist in physical and rehabilitation medicine. Some patients participated in a hybrid care model combining digital rehabilitation with in-person rehabilitation. 

The AI-based solution provided personalised exercises, telerehabilitation, regular reassessments, and continuous access to rehabilitation services. In the pilot, 145 patients started digital rehabilitation and completed an average of 3.5 exercise sessions per week. Patient satisfaction was high (9.2/10), pain intensity decreased by approximately 40% on average, and two-thirds of the patients achieved functional capacity without significant limitations. 

From an economic perspective, the cost of telerehabilitation was approximately EUR 275 per treatment episode, compared with EUR 307 for conventional contract-based rehabilitation, corresponding to a cost reduction of approximately 10%. Access to care also improved substantially, with the average waiting time decreasing from more than 60 days to 2.4 days. 

The study was an uncontrolled before-and-after pilot, and therefore no strong conclusions can be drawn regarding comparative effectiveness. The results suggest that AI-based telerehabilitation may be a feasible and cost-effective complementary option for selected patients with musculoskeletal disorders (MSDs), particularly by improving access to care through reducing logistical barriers. However, the digital solution was not sufficient for all patients, and some continued to require in-person rehabilitation. [3] 

The effects of several AI-based rehabilitation applications (Medico, medicalmotion, SelfBack) on pain, well-being, and self-management, as well as their safety, have been investigated in the literature. The applications generate personalised exercise programmes based on users’ health data, symptoms, and lifestyle factors. [4,5,6] AI-based exercise programmes were found to reduce pain to a clinically and statistically significant extent and to improve well-being in patients with back pain (p < 0.001 for both outcomes). [4,5] Functional limitations in patients with low back pain decreased significantly (p < 0.03) when the application was used as an adjunct to usual care. The effect was maintained at the nine-month follow-up. No differences between the groups were observed in quality of life, physical activity, or fear-avoidance beliefs. [5] The evidence also suggests that the AI system is capable of generating predominantly safe exercise recommendations, with 99.9% of the exercises assessed as presenting no risk. [6] 

At the time of the assessment, research evidence specifically related to the AIRe platform remains limited, and no peer-reviewed studies on the product have been published to date. The literature review suggests that AI-based rehabilitation solutions may improve access to care, support the delivery of rehabilitation, and be more cost-effective than usual care. However, the evidence is limited by heterogeneity across studies, short follow-up periods, and the small number of comparative studies. Further research is needed on long-term effects and cost-effectiveness.

Key aspects related to the safe use of the product have been identified and documented. According to the provider’s assessment, the product does not involve direct health risks. Potential risks are indirect in nature and mainly relate to incorrect use or use outside the intended purpose. Such risks may include, for example, misinterpretation of content, inappropriate self-directed actions, or use of information in ways for which the product was not intended. [1] 

The identified risks have been addressed through instructions for use, clear user interface design, and technical measures, such as functional restrictions and guiding mechanisms. [1] 

According to the provider, the product has not compromised client or patient safety. [1] 

There is no direct research evidence on the safety of the product. However, its safety has been developed, tested, and evaluated in research, development and innovation (RDI) projects in simulation and testing environments together with professionals and users. [1] 

The provider has a risk management plan in place that defines processes and reporting procedures for ensuring information security. A risk analysis has been conducted for the product, but it has limitations. The analysis focuses mainly on information security and does not cover key risks related to the product, its users, and the operating environment. At the time of the assessment, no documentation was available to demonstrate that the risk analysis is conducted and updated systematically. [1] 

The provider has identified risks associated with the use of artificial intelligence and has implemented measures to ensure that data stored in the product are logical, accurate, and consistent. This is ensured through an architecture based on defined data models, interface-specific validations, access control, and technical control and monitoring mechanisms. [1] 

The product takes into account age-specific recommendations used in Finland concerning health, functional capacity, and rehabilitation through targeted guidance at the levels of content, framework, and user interface. The health-related thresholds included in the product can be adjusted on an individual basis. [1] 

A person responsible for product safety has been designated. [1]

Costs are determined based on the service content and the number of users. [1]

The service organisation procuring the product determines the costs incurred by the end user for using the service. [1]

The initial implementation fee includes:

• • setting up the AIRe instance 
  • configuration 
  • definition of content 
  • user management 
  • integrations.

The use of the product involves ongoing costs, including: 

• • maintenance 
  • email support 
  • basic updates. [1] 

Costs depend on the language models used, service level, hosting solution, information security and data protection requirements, and integrations. Additional costs may arise from functional changes, new integrations, changes to the models used, or further development of the product. [1] 

As a browser-based solution, the AIRe platform generally does not require changes to physical facilities or equipment. However, its implementation may require changes to operational processes, instructions, access rights, and information security practices. The costs associated with these changes depend on the organisation’s internal processes. [1] 

User training is subject to a fee. The cost depends on the number of training days, the amount of user support required, the number of participants, the need for professional validation, and the scope of the training materials. [1] 

A calculated costsavings estimate has been prepared for the product, comparing its costs with a scenario in which an equivalent service is provided through appointments with healthcare professionals. [1] 

Based on the information provided by the manufacturer, the costs of using the service appear reasonable compared with providing an equivalent solution by other means.

However, there are some uncertainties in the cost estimates, as implementation and maintenance may vary between different operating environments. In addition, the product is still under development, and an established pricing model may not yet have been formed. [1]

The assessment was based on the information security and data protection requirements framework for social welfare and healthcare procurements [7] and the response material provided by AIRe. [1] 

Risk management and data security testing 

Key objectives have been defined for the product’s risk management, including data protection, ensuring continuity, and identifying and managing information security risks. Based on the description provided, the risk management process is systematic and covers risk identification, assessment, prioritisation, treatment, continuous monitoring, and updating. However, more detailed documentation on the proactive assessment and management of individual risks was not provided. [1] 

Information security is tested as part of the development process using measures such as automated dependency scanning, code reviews, and testing in a staging environment. A separate manual penetration test had not been conducted at the time of the assessment, but one is planned. [1] 

The information security and data protection risk management plan is updated regularly and whenever changes occur in the operating environment. [1] 

The provider has designated responsible persons for handling information security incidents. [1] 

Log management 

System activities generate log data, which are collected centrally using Azure Monitor. Based on the documentation, centralised monitoring covers service events, performance data, and error conditions. The collected data are used for analytics and monitoring system operation. [1] 

Based on the information available at the time of the assessment, monitoring is performed regularly, and serious incidents are addressed immediately. The architecture also supports alert-based monitoring. According to the provider, alerts can be implemented as an additional feature. [1]

User management 

The AIRe platform consists of two user interfaces: the AIRe Talk interface for clients and the AIRe Hub interface for professionals. User management is based on a centralised authentication and access management service and a role-based access control model that determines which functions and interfaces each user is authorised to access. Role-based permissions can be further refined with user-specific access rights when necessary. The system also supports application-specific and organisation-specific access restrictions. 

User account registration is based on an email address and password, with the email address verified using a code during registration. [1] Authentication and session management follow modern practices, and inactive sessions are automatically terminated after a defined period of inactivity. [1] 

The service supports two-factor authentication. [1] 

Hardware

The product is a Software as a Service (SaaS) solution implemented primarily using Microsoft Azure cloud services. The product also uses third-party services, including Azure OpenAI services. It is based on a modular service architecture in which different functionalities are divided into independent service components that communicate through application programming interfaces (APIs). [1] 

The service is provided as a browser-based web application. [1] 

Data protection 

According to the provider, no personal health data are stored in the system. Other data stored in the system are securely retained in the Azure cloud environment. The encryption methods and information security practices used are based on generally accepted and up-to-date standards. Data in transit are protected using the TLS encryption protocol, while data at rest are protected using the encryption and security mechanisms provided by the cloud service. [1] Passwords are handled using established hashing methods with user-specific salting and a defined number of iterations. [1] 

Data subject to the General Data Protection Regulation (GDPR) are stored within the EU and/or the European Economic Area (EEA). [1] 

A Data Protection Impact Assessment (DPIA) has been conducted for the product. [1] 

The provider acts as the data processor, while the customer organisation acts as the data controller. 

The company has appointed a Data Protection Officer (DPO). [1]

Other considerations 

The security of third-party services (OpenAI services) is partly based on assumptions made by the provider, and no separate risk analysis has been conducted for these services. Regarding the security of Artificial Intelligence (AI) and Large Language Models (LLMs), measures include ensuring the secure handling of inputs and prompts, considering the impact of access rights, minimising data, and restricting data processing. [1] 

General guidance on procurement  

During the procurement process, the organisation’s IT administration, information security specialist, and data protection specialist should always be consulted. The organisation should discuss with these experts whether the product meets its specific requirements. In addition, we recommend that wellbeing services counties use the European Union Agency for Cybersecurity (ENISA) cybersecurity guide for procurement as supporting guidance in their procurement processes. [8]

The AIRe Talk web application for clients is compatible with the most commonly used web browsers across different devices. A free demo version is available to clients as part of the implementation process. [1]

Usability 

The individuals who participated in the testing and development of the product represent actual end users. Approximately 350 people participated in the testing, including both rehabilitation clients and professionals from Finland, Rwanda, Kenya, Indonesia, and Vietnam. The participants ranged in age from approximately 18 to 80 years and varied considerably in terms of functional capacity, digital skills, and experience. Reports are available on both prototype testing and accessibility testing. [1,9,10,11,12] 

According to the provider, standards have not been applied in the design or testing of the user interface. Instead, the user interface has been developed in accordance with Human-Centered Design principles. [1] 

Usability testing has identified several areas for improvement, including small font size, insufficient contrast, lack of speech functionality, lengthy responses that may increase user burden, unclear aspects of the user interface, and limitations in keyboard and screen reader support. The provider will take these findings into account in further product  development. The provider has recognised that use of the product may be challenging for people with visual, hearing, cognitive, or fine motor limitations, although these limitations do not prevent use of the product. End users can provide feedback and development suggestions regarding the usability and accessibility of the product by email. The provider takes this feedback into account in the further development of the service. [1] 

Accessibility 

The textual content of the product has been designed to use clear and understandable language. [1] 

According to the provider, the following user groups have been taken into account in the design of the product’s usability and accessibility: 

• • People with low vision or visual impairments: options to adjust font size and colour theme, as well as support for voice functionalities (speech-to-text and text-to-speech). 

• • People with colour vision deficiencies: option to adjust the colour theme. 

• • People who benefit from clear communication: interaction adapted to the user’s needs and a system prompt that guides the AI to use clear language and concise expressions. 

• • People with cognitive difficulties (e.g. perception, memory, or learning): reminder functionality; a system that learns about the user as the conversation progresses; a clear user interface; and a system prompt that guides the AI towards understandable interaction, including short responses, one question at a time, and clear language. 

• • People with fine motor difficulties: support for voice functionalities (speech-to-text and text-to-speech) and the option to scale up the user interface. 

The following user groups have not been specifically considered in the design, but the platform can be used by them:

• • People with hearing impairments and sign language users: text-based interaction is available. 

• • People with epilepsy: a calm visual design with no flashing content or automatically playing videos, and limited use of animations and moving content. 

An accessibility assessment of the product has been conducted by an external party. The product partially complies with the technical requirements of the European standard EN 301 549. More detailed information is provided in the accessibility statement, which is available to registered users under the About AIRe Service menu. Accessibility feedback can be submitted by email. According to the provider, feedback is responded to within 14 days. [1,14] 

The instructions for use are accessible and available upon request from AIRe product support. For developers, the instructions are also publicly available on GitHub. [1,15] 

General guidance on procurement  

The procuring organisation should take into account the accessibility requirements set out in the legislation in force at the time of procurement. [16,17,18]

Interoperability 

The product can be integrated with other software systems, such as electronic health record systems or Kanta Services, if required. However, such integrations are not planned for the current version of the product. [1] 

Data can be transferred using standardised file formats, such as JSON. Some data are stored and transferred in their original format, and the related specifications are publicly available on GitHub. [1,15]

Technical stability 

All source code and configuration files are subject to internal version control and audit processes. Testing processes have been defined for the product and are carried out as part of product development. The provider receives and processes error reports submitted by end users through JAMK product support. When necessary, error reports are forwarded to the technology partner. It is possible to roll back to a previous version of the product. [1] 

The provider proactively monitors system operation by tracking response times, error conditions, and token consumption to identify potential faults. A project plan has been prepared for the decommissioning of the product. [1] 

According to the manufacturer, the product has experienced a few service interruptions lasting several hours during the past six months due to outages affecting third-party services. [1] 

Updates to the software libraries used by the product are carried out as part of normal development activities and are included in the testing process. Updates may cause service interruptions, but users are notified through the AIRe service two business days in advance and during the maintenance period. The product requires an internet connection to function properly. [1] 

Observations about artificial intelligence (AI) 

AIRe is a platform solution developed by JAMK University of Applied Sciences. The product is not dependent on a single AI model; instead, models can be replaced through compatible interfaces. The current version of the product uses Microsoft Azure OpenAI models and the Azure cloud platform. According to the provider, external AI is used through the AIRe platform in a limited manner and for specific purposes. Its use is controlled through configurations, theme-specific prompts, trusted sources, and professional oversight. [1] 

According to the provider, the system currently falls within the limited-risk category under the EU AI Act, as it operates as a transparent, professionally supervised interactive support system without autonomous decision-making. The provider partially monitors potential risks associated with the AI solution. [1] 

JAMK, as the owner of the platform, is responsible for the further development, updates, and decommissioning of the AI system together with the technology developer. External organisations may also deploy AIRe, in which case the organisation implementing the specific instance is responsible for its updates and potential decommissioning together with the technical provider. 

The system stores log data on its operation and use of data. The logs contain information such as: 

• • users’ visits and activities within the service, 

• • system performance, response times, and latency, and 

• • error conditions. 

The AI system enables human oversight and control to ensure its safety and reliability. This is implemented by professionals defining the themes, sources, prompts, and content that guide the system’s operation and monitoring its performance. Based on feedback and evaluation, professionals can modify, restrict, or suspend the operation of the AI system. [1] 

Documentation is available for the AI system, in which the training data, training methods, and outputs generated by the model are partially described. In the professional user interface, AIRe Hub, the prompts, sources, documentation, themes, and settings that guide the behaviour of the system (AIRe Talk) are visible depending on the user’s access rights. However, the actual training data and training methods of the AI model used are covered by the documentation provided by the language model provider. [1] 

According to the provider, the performance and functionality of the AI system have been tested to a limited extent under controlled conditions and in selected pilots under conditions corresponding to the actual operating environment. 

The functionality of the AI system has been verified in the following exceptional situations: 

• • Incomplete data: AIRe requests additional information from the user and prompts the user to limit conclusions to the defined intended purpose. 

• • Incorrect user inputs: Professionals guide AIRe’s interaction through prompts and by defining the scope of themes. When necessary, the user is redirected to a professional or another service. 

• • Changes in the operating environment: Configurations, content, sources, prompts, and, when necessary, the underlying model can be updated in the background without rebuilding the entire platform. 

• • Attacks and manipulation: Risks are mitigated through strict access controls, restricted integrations, trusted sources, logging, provider-specific security practices, and controlled change management. 

Other unexpected situations may arise from factors such as language variations, hallucinations, excessive user reliance on the system, and third-party service outages. These risks are addressed through measures including a clearly defined intended purpose and restricted functionalities, user guidance, and human oversight. 

The AI system includes safeguards designed to prevent responses that violate defined rules and to ensure that the AI operates within its intended purpose. 

According to the provider, measures have been taken to ensure that the principles of equality and non-discrimination are upheld in AI-supported decision-making, as AIRe does not make autonomous decisions at the platform level, for example, regarding access to services. The risk of discrimination is also mitigated through professionally defined content, sources, and guidance logic, as well as by validating the solution for specific target groups and contexts. [1] 

End users are informed that they are interacting with an AI system at the beginning of the conversation, within the user interface, and through the terms of use and user instructions. [1]

Data Sources 

The AI model uses data from the following sources: information and conversational inputs provided by end users; questionnaire responses; content and sources added by professionals; and third-party services connected through interfaces. However, the sources used always depend on the specific use case. [1] 

The AI solution does not require a specific number of input variables, as the required inputs depend on the intended purpose, the course of the conversation, the selected themes, and any integrations used. [1] 

No single quantity of training and validation data can be specified for the platform used by the AI solution, as AIRe is not based solely on a single model trained by the provider itself. [1] 

According to the provider, efforts have been made to ensure that the data sources used for training and validation are relevant and unbiased with respect to the actual use case. This is supported by professionally curated content, a clearly defined intended purpose, validation specific to the operating environment and/or intended purpose, and the fact that AIRe does not rely solely on a single general information source. [1] 

According to the provider, appropriate rights to use the data have been ensured throughout all AI training and validation phases. These rights are verified separately for each deployment. This applies to the model provider’s terms as well as the legal basis for using content, sources, interfaces, and personal data incorporated into the AIRe platform. [1] 

According to the provider, data protection and information security have been taken into account throughout all phases of AI model training and validation. GDPR principles have been incorporated into the development and use of the platform from the outset. This is ensured through measures including data minimisation; role-based access controls; encrypted data transmission; hosting servers in Northern Europe; and anonymisation of data whenever possible.

System Retraining 

The AI model is not a continuously learning system in production use. New data are only utilised in separate, controlled further development and use-case-specific improvements. The provider has ensured the relevance of data used for retraining by not automatically using user inputs or model outputs for this purpose. Data validated by professionals are also separately approved through a controlled process. [1] 

The provider has also taken measures to ensure that biased or erroneous outputs generated by the system do not influence future inputs through feedback loops, as AIRe outputs are not automatically fed back into the training data. Any further development is based on separate evaluation, approval, and documented change management. [1] 

AI Performance Parameters 

At the time of the assessment, information on the performance parameters of the AI solution was not available. According to the provider, common AI performance metrics, such as accuracy, precision, and recall, are not well suited to evaluating AIRe. This is because AIRe does not make diagnoses, classify patients, or generate predictions. Instead, AIRe functions as a guidance, support, and interactive service, where the emphasis is on usability, safe guidance, human oversight, task-specific performance, and impact measures related to implementation. [1] 

The provider informs organisations considering procurement about the performance of the AI through technical documentation, instructions for use, demonstrations, pilot results, and use-case-specific validation reports. The instructions for use of the AI solution also partially describe its performance. [1] 

Decision-Making, Explainability and Transparency of the Model 

When using the product, it should be taken into account that the role of professionals may increasingly shift towards defining system parameters, supervision, and managing complex situations, as the product enables a digital-first operating model. In this model, some routine guidance, counselling, support for goal setting, monitoring, and navigation to appropriate services are provided digitally within the content, guidance, and limitations defined by professionals. [1] 

In the decision-making process: 

• The human professional defines the boundaries of use, approves the content and sources, oversees the system’s operation, and remains responsible for the final professional judgement. 

• The role of AI is to support interaction, suggest content, structure information, and support progress monitoring. 

The rationale and uncertainties associated with the AI model’s decisions are communicated to users through instructions for use, onboarding, professional training, a clearly defined intended purpose, and documentation. Responsibility for training lies with the implementing organisation together with the platform provider or administrator. [1]

Training and product support 

The provider offers product support by email on weekdays from 10:00 to 14:00. It also provides training in Finnish and English, organised in connection with implementation and tailored collaboratively to the specific organisation or pilot. AIRe Talk and AIRe Hub user instructions are available to the customer organisation to support implementation. The product’s instructions for use are available in Finnish. [1] 

Consumer Protection 

The provider offers support for user enquiries on weekdays from 10:00 to 14:00. 

The intended purpose of the product is communicated to end users on the website and in AIRe Talk.

Users can familiarise themselves with the product before implementation upon request. The provider also aims to launch an open trial environment during autumn 2026. [1] 

Exact pricing information cannot be provided prior to implementation, as pricing is determined on an organisation-specific basis. The product does not include advertising, promotional offers, automatically renewing subscriptions, or limited-time special offers. [1] 

If a user wishes to discontinue use during the care process, they can request that their account be closed or, alternatively, delete the account themselves and anonymise the associated data. [1] 

Achieving the intended health benefits requires support from a healthcare professional. This is communicated to end users in the implementation information and service guidance. [1] 

Age restrictions apply to the use of the product. An age limit of 18 years has been applied in the pilots. For other use cases, the age limit can be defined separately. [1] 

The provider has taken into account the suitability of the content and instructions for different age groups. Professionals define the content, themes, and guidance according to the specific target group. [1] 

Distribution of the product 

The product is used at JAMK University of Applied Sciences as a simulation and testing environment for research, development and innovation (RDI) activities and education. In addition to Finland, testing and co-development activities have been conducted in Vietnam, Indonesia, Rwanda, and Kenya. [1] 

1. The Digi-HTA questionnaire filled out by the company, not public information.
2. Gebrye, T., Mbada, C., Niyi-Odumosu, F., Fatoye, C., Odetunde, M., Hakimi, et al. (2026). Cost-effectiveness of artificial intelligence interventions for musculoskeletal disorders of the spine: A systematic review. Rheumatology International, 46(6), 107. https://doi.org/10.1007/s00296-026-06122-3 
3. Branco, J. P., Graça, D. T., Costa, E., Branco, C. A., Sampaio, F., et al. (2026). AI based rehabilitation: The way forward in addressing unmet needs in musculoskeletal disease. Frontiers in Public Health, 14, 1773733. https://doi.org/10.3389/fpubh.2026.1773733 
4. Griefahn, A., Avermann, F., Zalpour, C., Marshall, R. P., Cordon Morillas, I., & Luedtke, K. (2025). Exploring the effect of an 8-week AI-composed exercise program on pain intensity and well-being in patients with spinal pain: Retrospective cohort analysis. JMIR Formative Research, 9, e57826. https://doi.org/10.2196/57826 
5. Sandal, L. F., Bach, K., Øverås, C. K., Svendsen, M. J., Dalager, T., Stejnicher Drongstrup Jensen, J., et al. (2021). Effectiveness of app-delivered, tailored self-management support for adults with lower back pain-related disability: A selfBACK randomized clinical trial. JAMA Internal Medicine, 181(10), 1288–1296. https://doi.org/10.1001/jamainternmed.2021.4097 
6. Griefahn, A., Zalpour, C., & Luedtke, K. (2024). Identifying the risk of exercises, recommended by an artificial intelligence for patients with musculoskeletal disorders. Scientific Reports, 14, 14472. https://doi.org/10.1038/s41598-024-65016-1 
7. National Cyber Security Centre Finland. Information security and data protection requirements for social welfare and healthcare procurements. Available at: https://www.kyberturvallisuuskeskus.fi/fi/ajankohtaista/ohjeet-ja-oppaat/sosiaali-ja-terveydenhuollon-hankintojen-tietoturva-ja. Accessed 13 August 2026. 
8. European Union Agency for Cybersecurity (ENISA). Procurement Guidelines for Cybersecurity in Hospitals. Available at: https://www.enisa.europa.eu/publications/good-practices-for-the-security-of-healthcare-services. Accessed 13 August 2026. 
9. Eriksen, M. Designing an intelligent Digital Rehabilitation Platform through collaboration with service users and professionals. JAMK University of Applied Sciences. Available at: https://www.jamk.fi/en/article/designing-an-intelligent-digital-rehabilitation-platform-through-collaboration-with-service-users. Accessed 4 August 2026. 
10. Eriksen, M. (2024). Kuntoutuksen tekoälyn ensimmäiset askeleet maailmalla [First steps of rehabilitation AI around the world]. JAMK Arena Public. https://urn.fi/urn:nbn:fi:jamk-issn-2984-0791-102 
11. Dewi, R. S., Widyaningrum, R., Pratiwi, A. A., & Sholihah, M. (2024). AIRe platform: Indonesian insights in developing AI-driven rehabilitation solution. In J. Chichaeva (Ed.), Global Digital Rehabilitation Summit 2024: Abstract book (p. 14). JAMK University of Applied Sciences.https://www.theseus.fi/bitstream/handle/10024/877270/GDRS2024_Abstract_book-FINAL.pdf?sequence=2
12. Huotari, J., Heiskanen, K., Mälkönen, I., Eriksen, M. M., Murtonen, K.-P., & Nyman, H. (2025). AIRe platform: Expanding global access to multidisciplinary rehabilitation through AI-driven innovation and technology [Platform presentation]. World Physiotherapy Congress 2025, Tokyo, Japan. https://doi.org/10.82161/aa5x-wg89 
13. Oduor, M., Lällä, K., Eriksen, M., Korniloff, K., & Murtonen, K.-P. (2025, October 7). Bridging AI and primary care: AI-assisted health system integration model for the Inclusion App. JAMK Arena. Available at: https://arena.jamk.fi/en/arena-pro/bridging-ai-and-primary-care-ai-assisted-health-system-integration-model-for-the-inclusion-app/. Accessed 10 August 2026. 
14. JAMK University of Applied Sciences. (2026). Tervetuloa AIRe-palveluun [Welcome to the AIRe service]. Available at: https://talk.aire.fi/. Accessed 5 August 2026. 
15. GitHub. (2026). AIRe Platform. Available at: https://github.com/AIRe-Platform. Accessed 5 August 2026. 
16. Act on the Provision of Digital Services 306/2019 (Laki digitaalisten palvelujen tarjoamisesta 306/2019). Available at: https://www.finlex.fi/fi/laki/alkup/2019/20190306. Accessed 14 August 2026. 
17. Finnish Transport and Communications Agency Traficom, Supervision of Digital Accessibility. Requirements of the Act on the Provision of Digital Services for service providers. Available at: https://saavutettavuusvaatimukset.fi/fi/digipalvelulain-vaatimukset/digipalvelulain-vaatimukset-toimijoille. Accessed 14 August 2026. 
18. Act on Public Procurement and Concession Contracts 1397/2016 (Laki julkisista hankinnoista ja käyttöoikeussopimuksista 1397/2016). Available at: https://www.finlex.fi/fi/laki/alkup/2016/20161397. Accessed 14 August 2026. 

Jari Haverinen, Senior Planning Officer, FinCCHTA 

Raija Järvinen, Senior Planning Officer, FinCCHTA 

Teemu Mustola, Senior Planning Officer, FinCCHTA