BIOMECÁNICAMENTE

Revista de Biomecánica
Reading time 1 min.
  Play audio




Entrada anterior
Older people and ageing
        

AI3cord: IoT devices and Artificial Intelligence to monitor and prevent frailty in older people. Featured

Older people and ageing

26 December 2023.

 

 

 

Author(s): Arturo Gómez Pellín, Sergio Jiménez Monferrer, José Francisco Pedrero Sánchez, María Amparo López Vicente, José David Garrido Jaén, Carlos Atienza Vicente.

 

The high social and economic cost of a progressively ageing global population highlights the need to promote effective models of healthy ageing that help to prevent or delay disease.

In this context, the Instituto de Biomecánica (IBV), through the Strategic Cooperation Projects programme of the Valencian Innovation Agency (AVI), has set up a consortium of Valencian stakeholders from the technology and healthcare sectors to carry out the AI3cord project, a cutting-edge R&D&I project for the development and evaluation of an innovative technological solution for the monitoring and prevention of frailty and chronicity in the elderly.

AI3cord has made it possible to create a technological platform that uses IoT devices and AI-based data analysis techniques to remotely and dynamically evaluate new indicators aimed at monitoring and preventing frailty in the elderly, with the least possible intrusion.

This project opens the door to its application in the treatment of other pathologies or chronic ailments that can benefit from remote and non-invasive monitoring technology.

 

 

INTRODUCTION

There is a strong trend towards the progressive ageing of the world’s population that entails significant social and economic costs. In addition, this situation is particularly exacerbated by the impact associated with the illnesses suffered by this highly dependent segment of the population. This context highlights the need to transform current social and healthcare policies, which are based on interventionist, reactive and disease-focused models, towards more sustainable models based on prevention and health maintenance.

It is therefore crucial to understand how we can effectively promote a model of healthy ageing that prevents and delays the onset of disease and the loss of capabilities and autonomy that leads to early dependency.

The magnitude and relevance of this problem is highlighted in the Smart Specialization Strategy for Research and Innovation in the Autonomous Community of Valencia (RIS3-CV), which has been designed to concentrate intensive R&D&I resources that exist to drive the economy and transformation of our community’s production model. This strategy has defined active and healthy ageing and the improvement of care for people in a situation of fragility or chronic illness through innovative technological solutions as one of the 4 pillars of R&D&I specialization in the Autonomous Community of Valencia and as a priority objective in the field of health.

Similarly, the Strategic Innovation Committee Specialized in Health, an advisory body of the Valencian Innovation Agency (AVI) whose mission is to provide innovative technological solutions to the priority challenges of the Autonomous Community of Valencia in the field of health and to accelerate their implementation in the business community, has identified the control and prevention of chronicity and frailty as one of the 3 main priority challenges.

In this context, the Instituto de Biomecánica (IBV), through the Strategic Cooperation Projects programme of the (AVI)[1], has set up a consortium of Valencian stakeholders from the technology and healthcare sectors to carry out AI3cord (Figure 1), a cutting-edge R&D&I project for the development and evaluation of an innovative technological solution for the monitoring and prevention of frailty and chronicity in the elderly.

 

 

Figure 1 . Partners in the AI3cord project

AI3cord has gathered the knowledge and skills of MySphera, a leading European technology SME that develops smart sensors and has deployed solutions based on the Internet of Things (IoT) in more than 50 hospitals, and S2 Grupo, a large company that invests heavily in R&D&I and is a leader in cybersecure data management solutions, event monitoring and collection and advanced intelligence. In terms of clinical care, the consortium has benefited from the significant contribution of the Research Foundation of the General University Hospital of Valencia (FIHGUV), a research organization with extensive innovation capacity and highly experienced in the clinical assessment of older people and active ageing technologies.

The AI3cord initiative has combined the experience of its partners and the latest advances in technologies such as (i) Artificial Intelligence (AI), (ii) Big Data processing, (iii) IoT devices for activity monitoring and vital signs measurement, and (iv) 5G systems for high speed transfer of massive amounts of data, in order to develop an innovative technology capable of remotely, dynamically and proactively assessing indicators of the health status of older people and of detecting risk situations, through the acquisition and analysis of data recorded in a non-invasive manner during their daily activity in their own environment.

This solution is aligned with a “liquid hospital” type of care model, which is based on the digitalization of processes to care for patients beyond the physical boundaries of healthcare centers, using flexible technological solutions adapted to their environment and needs, in this case telemonitoring.

In this way, AI3cord aims to respond to the needs raised by the main healthcare and research strategies by providing a solution for the monitoring and prevention of frailty and chronicity in the elderly, based on individualized, preventive, de-localized medicine and focused on maximizing the value of care.

 

DEVELOPMENT 

The AI3cord project included 3 phases: 1) platform design and development, 2) demonstration study and 3) processing, data analysis and evaluation of the user experience.

PHASE 1 – Platform design and development

During the solution design phase, we identified six domains of frailty[2] as relevant for their potential to be assessed in a remote and non-invasive manner using technological devices (Figure 2). We excluded other domains that, they can only be assessed in clinical care settings and/or are not suitable for remote and non-invasive recording (e.g. genetic factors, hormonal markers, inflammatory markers, etc.).

Figure 2. Classification of fragility domains and related indicators based on various studies and projects that were consulted, aimed at assessing fragitliy remotely using technology.

 

We selected the following domains: physical activity, rest, cognitive decline, nutritional status, social situation and emotional state. For each of these domains, we identified heterogeneous indicators that have shown to be highly correlated with the degree of frailty in older people and that could be recorded using remote and non-invasive monitoring technologies. To complement the above indicators, we also identified the most valuable socio-demographic and clinical data in relation to an individual’s frailty status.

The identification of these indicators was facilitated by the work of a number of healthcare professionals who took part in discussion groups and in-depth interviews to analyse their experience and needs in relation to monitoring the health and frailty status of older people, and to define indicators aimed at preventing the loss of capabilities associated with frailty and chronicity, and at the early detection of risks requiring intervention.

We then conducted an in-depth analysis of technologies capable of monitoring, in a remote and non-invasive manner, the defined indicators and that met the following requirements, among others: 1) integration into the individual’s home environment and minimal intrusion into an individual’s daily routine, 2) accurate and reliable measurements, 3) ease of use by older people lacking technical expertise, 4) compatibility for real-time monitoring of information for anomaly detection and early intervention, 5) protection of individual privacy and confidentiality, 6) compatibility for easy integration and data exchange, and 7) cost-effectiveness so as not to place an excessive economic burden on older people or their carers.

Based on this work and on the extensive experience of the IBV and the other partners in active ageing and health assessment technologies, we defined the specifications of the AI3cord platform both in terms of its technical aspects, such as the type of data, frequency of measurements, connectivity, autonomy, maintenance, etc., and its non-technical aspects, such as usability, security, legal considerations, accessibility, adaptability to the environment, etc. Finally, on the basis of the defined specifications and after evaluating different proposals and defining the most appropriate architecture for the network of sensorization, communication, data processing and storage devices, we selected the most appropriate devices. para dar una visión general

Figure 3. Diagram of the AI3cord platform

Figure 3 gives a general overview of the AI3cord platform, which consists of:

– A network of IoT devices that collects information both in an individual’s home and in their external environment during their daily activity.

– A gateway that connects the IoT devices, receives and distributes the data, handles any communication losses, and encrypts and secures the communications.

– A central server whose functions include the collection and storage of the collected data in a database, the preliminary analysis of this data, the deployment of the various microservices that are responsible for the connection with the different data sources, the use of APIs for the integration of the applications and the provision of data for processing by the AI modules.

– Two applications with their respective interfaces (dashboards) for the recording and visualization of information both by the user him or herself (mobile application, Figure 4) and by the clinician (web application, Figure 5).

Figure 4. AI3cord  mobile app

 

Figure 5. Patient data recording and visualization interface for clinical use.

 

Figure 6 shows the specific IoT devices used in the AI3cord platform to monitor the health status of the individuals at home and in their outdoor environment:

– A smart watch that measures parameters such as heart rate, walking speed, number of steps, burned calories and distance travelled.

– A mobile application that records, by means of regular questionnaires, aspects related to the individual’s perception of loneliness, fatigue, sleep quality or memory, among others.

– Presence sensors that detect the individual’s activity at home and assess aspects related to their physical activity, as well as relevant behavior patterns in frailty situations (e.g., feeding or grooming routines).

– A door sensor to detect when the individual enters and leaves their home and assesses aspects such as time spent at home or away from home, or time spent at home with someone else.

– Digital smart scales for regular weight monitoring.

Figura 6 . Dispositivos de la plataforma AI3cord

 

PHASE 2 – Demonstration study

During the second phase of the project, we conducted a demonstration study to evaluate the use of the AI3cord platform in a real-world setting, with a sample of 23 people over the age of 60 in different stages of frailty.

The study, which lasted 8 months, allowed us to continuously collect data on the participants’ physical, cognitive, social, emotional, nutritional and rest status, through the AI3cord platform.

The participants were also asked to go the General Hospital on two occasions (at the beginning and at the end of the study) to be assessed by the project’s clinical staff using various questionnaires and standardized clinical tests in order to characterize their health status and degree of frailty from a clinical point of view.

Figure 7 shows a schematic of the described pilot study methodology.

Figure 7. Pilot study methodology

 

PHASE 3 – Processing, data analysis and evaluation of the user experience

During the final phase of the project, we processed and analyzed the data recorded during the study and we assessed the participants’ experience of the platform.

The following is a summary of the main analyses carried out and the results obtained.

 

Voice-based assessment of emotional state

In addition to regularly completing questionnaires related to frailty factors through the mobile application, users were able to register how they felt during the week via voice recording (Figure 8).

Figure 8. Functionality of the AI3cord app for voice recording mood.

 

The recorded audio files were then analyzed using natural language processing techniques and AI-based algorithms to estimate the emotional state of the participants during the study, based on the language they used (Figure 9).

 

Figure 9. Analysis process for voice-based emotional state assessment.

 

Detection and analysis of anomalies through the smartwatch

The aim of this analysis was to detect any anomalous patterns in the signals received from the smartwatch (i.e., heart rate signal, number of steps and walking speed) through the combined use of statistical techniques or neural networks (Figure 10).

Figure 10: Example of the heart rate signal of an AI3cord participant using a SARIMA model. X-axis: Time, Y-axis: Mean BPM per 30 minutes (blue), SARIMA prediction (orange) and SARIMA confidence interval (shaded). Detected anomalies are shown as vertical lines

 

The application of these techniques made it possible to detect significant changes in the data recorded by the smartwatch that were associated with the onset of clinical events recorded during the study, such as physical health problems, medical procedures or a deterioration in the participants’ mental health.

Group classification and follow-up

The aim of this analysis was to use clustering techniques to create groups of people with similar characteristics according to multiple frailty-related indicators recorded during the study using the AI3cord platform.

These clusters make it possible to track an individual’s status over time and to send an alert if they move away from their group and are at risk of joining another group with a higher level of frailty, so that preventive therapeutic measures can be taken.

Evaluation of the user experience

In order to evaluate the users’ experience of the platform by both the healthcare professionals and the elderly participants in the study, we carried out in-depth interviews to gather their opinions on, among other things, the ease of use and level of acceptance of the platform, the interpretation of the results, the quality and usefulness of the information provided, the extent to which it had met their expectations, and the identification of shortcomings and unmet needs, etc.

In general, the AI3cord platform was rated very positively, with users highlighting, among other things, the fact that it improves the problem it addresses, is convenient and easy to use, and offers great potential for monitoring frail people.

 

CONCLUSIONS

 

The AI3cord project has made it possible to create a technological platform that uses IoT devices and AI-based data analysis techniques to remotely and dynamically evaluate new indicators aimed at monitoring and preventing frailty in the elderly, with the least possible intrusion.

The study has made it possible to evaluate the use of the AI3cord solution in real-world settings and the extent to which the recorded indicators are of value in detecting changes in people’s health.

The evaluation of the participants’ user experience has shown a high level of acceptance, participation and satisfaction with the AI3cord technology.

The AI3cord project, thanks to the flexible and extensible nature of its open IoT architecture, opens the door to its application in the treatment of other pathologies or chronic ailments that can benefit from remote and non-invasive monitoring technology, such as advanced heart failure[1] or rehabilitation in rural areas[2], fields in which the IBV is actively working.

 

ACKNOWLEDGEMENTS

AI3cord has been carried out thanks to the support and funding of the Valencian Innovation Agency through its 2021-2023 programme to support the development of major R&D&I projects in collaboration with stakeholders in the Valencian Innovation System (INNEST/2021/237), and to the collaboration of the four partners in the project: MySphera, S2 Grupo, FIHGUV and the IBV.

 

[1] The AI3cord project falls within the framework of Challenge 1 of the Strategic Innovation Committee Specialised in Health, aimed at the control and prevention of chronicity and frailty.

[2] La multidimensionalidad de la fragilidad hace referencia a las diferentes características y aspectos de las personas y su entorno que pueden condicionar o verse afectadas con el declive de la salud y las capacidades.

[3] PCP Tique – https://www.tiquepcp.eu/

[4] PCP Rosia – https://rosia-pcp.eu/

 

AUTHOR’S AFFILIATION

Instituto de Biomecánica de Valencia
Universitat Politècnica de València
Edificio 9C. Camino de Vera s/n
(46022) Valencia. Spain

 

 

Volver al principio
Entrada siguiente