This research proposes an intelligent empathic robotic system for elderly healthcare, combining AI, Theory of Mind, and Afective Computing to deliver personalized care. The low-cost robot continuously monitors patients' physical and emotional states through wearable data and neural networks, ofering tailored support, therapy management, and emergency responses. By integrating ethical principles and patient-specific profiles, the system fosters trust and improves communication between patients, caregivers, and providers. This approach aims to enhance healthcare eficiency, reduce costs, and improve patient well-being through continuous, personalized interaction.
CEUR ceur-ws.org
The current healthcare technologies designed for elderly patients often do not provide the level of
personalized interaction necessary for efective care. Existing eHealth solutions are not integrated into
the care processes and value chains, often function as stand-alone add-ons to standard treatment rather
than supporting patients efectively with their self-care [
Many traditional healthcare services do not consider the human factor and are not adapted to each
user’s individual needs and capabilities [
Another significant gap in existing healthcare technologies is the lack of attention to the emotional
well-being of patients [
Efective healthcare for elderly patients requires continuous and comprehensive monitoring of their health status [11]. The lack of continuous monitoring of the health status of elderly patients is a significant gap that needs to be addressed through comprehensive geriatric evaluations, primary care involvement, and community involvement to efectively manage frailty and improve results for the aging population. This insuficiency may cause delayed reactions to medical problems, occasionally with dire repercussions.
The current healthcare system often imposes high costs on elderly patients, driven by frequent and sometimes unnecessary hospital visits and diagnostic tests [12, 13]. These costs can be a significant burden, especially for those on fixed incomes. By leveraging technology to provide more eficient and cost-efective care, the financial burden on elderly patients can be significantly alleviated.
The proposed system aims to improve patient well-being, optimize the organization of medical work, and reduce healthcare costs through more eficient resource management. In addition, it aims to counteract the unreliable self-diagnosis prevalent on the Internet by providing clear and personalized information and explanations. Once fully developed, the proposed system would bring significant savings to National Health Systems regarding: • Reducing unnecessary tests, specialist visits, and emergency room visits; • Conversely, decreasing hospital overcrowding, hospital days, and hospital costs related to patients delaying consultation with a specialist or going to the emergency room; • Reducing the workload and stress for doctors and caregivers.
The agent will interact with the patient based on a user profile that will include at least the patient’s medical history, preferences, and ethical principles. It will also interact with doctors and caregivers to provide them with feedback on the patient and to obtain advice and assistance when necessary. The focus will be on aspects related to the ethics of agent-patient interaction and the trust relationship that should ideally form.
The methodology involves designing an intelligent agent using logical programming to ensure compliance with safety and ethical standards. The Theory of Mind [14] and Afective Computing [ 15] will be integrated to enable the agent to understand and respond to patients’ emotional and cognitive states. The use of detailed patient profiles, including preferences, medical histories, and ethical considerations, will allow for personalized interactions and adequate support. The platform will be tested on real case studies, with the possibility of continuous refinement of the theory and implementation.
The next phase involves the integration of the intelligent agent with wearable devices [16, 17] that will continuously monitor various health parameters of the patients. The data collected from these devices will be analyzed by the agent to assess the health status of patients in real time. By providing continuous and comprehensive monitoring, the agent can detect early signs of potential health problems and alert healthcare providers or caregivers promptly.
To enable a natural and efective interaction with patients, the intelligent agent is installed on a robotic platform. This platform will serve as a physical embodiment of the agent, enabling it to communicate with patients through speech, gestures and other forms of non-verbal communication. We intend to install the agent on a robot because it has been seen that human users prefer to interact with a humanoid robot rather than with a computer program. Also, a robot is equipped with devices that can detect patient emotional signs that are useful for AF. So, the agent will be able to unobtrusively monitor the patient’s health and emotional state, arrange therapies, keep doctors informed, and request the intervention of the physician or caregiver when necessary.
The initial step toward achieving these objectives is to develop ”Blueprint Personas” [18] , a tool designed to foster person-centered care. This approach identifies patient profiles based on diverse needs, environments, especially within home settings, and a range of health and socioeconomic characteristics. These personas will be created from a questionnaire survay and will serve as detailed representations of diferent types of users, based on data collected from real patients. We will consider the pathology of Chronic Obstructive Pulmonary Disease (COPD), also taking into account doctors’ expectations for the agent.The questionnaire for our research will be structured into several sections to ensure a complete collection of the necessary information. Each section is designed to capture diferent dimensions of the participants’ experiences and needs.
Demographic Data: This section aims to understand the environmental context in which the individuals live. Questions in this category collect information on age, gender, occupation, education level, and geographic location. This demographic information provides a foundational understanding of the population being studied and helps to identify patterns and trends between diferent groups.
Habits: This category includes questions about the participants’ daily routines, dietary habits, and physical activities. Understanding their habits allows for a deeper insight into their lifestyle choices and how these may impact their health and well-being. For example, questions about the type of diet followed and the frequency of sports or exercise can reveal important aspects of their health behaviors.
Pathology and Technological Support: Participants are asked about their specific health conditions and how technology or technological support could help them manage their disease. The questions are designed to explore their experiences with their pathology, the challenges they face, and their openness to using technological solutions. This information is crucial for identifying the needs and preferences for technological interventions.
Expectations from the Intelligent Agent: This includes questions on the functionalities they deem essential, their desired level of interaction with the agent, and any specific features they believe would significantly aid them in managing their health. Understanding these expectations helps in tailoring the intelligent agent to better meet the users’ needs and enhance their overall experience.
This will ensure that the personas are complete and accurately represent the unique requirements and interactions relevant to the management of COPD. The creation and use of these personas and ontologies will significantly improve the personalized interaction capabilities of the intelligent agent.
As the next step in this project, we will design a complete architecture for the intelligent agent. This architecture will integrate various components to ensure that the agent can interact efectively with patients and doctors, personalize its responses, and provide reliable support. The architecture will include ontologies to represent ethical and trust aspects and may include a Knowledge Graph to contain the user profile and interaction history with the patient.
To ensure the practical applicability and efectiveness of the intelligent agent, it is essential to test and refine its capabilities through real-world case studies. In these studies, the agent is used in real environments in which users interact with it as part of their daily lives. By observing and analyzing these interactions, we can identify potential areas for improvement, uncover unforeseen challenges and gain valuable insights into the user experience.
Trust is a crucial factor for the adoption and sustainable use of technological solutions in healthcare. The agent must operate transparently and provide clear and understandable explanations for its recommendations and actions. Privacy and data security are also crucial. Users must be assured that their personal and medical information will be treated with the utmost care and confidentiality. In addition, the agent should be guided by ethical principles in its interactions and ensure that it supports users in a respectful and non-intrusive manner.
This research project has the potential to revolutionize home healthcare for all chronic patients, particularly elderly adults with chronic diseases. By combining artificial intelligence, robotics, and medical expertise, this initiative ofers a promising solution to a more sustainable and efective healthcare system.
Research partially supported by the PNRR Project CUP E13C24000430006“Enhanced Network of intelligent Agents for Building Livable Environments - ENABLE”, and by PRIN 2022 CUP E53D23007850001 Project TrustPACTX - Design of the Hybrid Society Humans-Autonomous Systems: Architecture, Trustworthiness, Trust, EthiCs, and EXplainability (the case of Patient Care), and by PRIN PNNR CUP E53D23016270001 ADVISOR - ADaptiVe legIble robotS for trustwORthy health coaching.
I gratefully acknowledge the constant support and guidance of my supervisor, Stefania Costantini, whose encouragement and insights were invaluable throughout this research. [10] V. Espejo-Fernández, P. Martínez-Angulo, Psychosocial and emotional management of work experience in palliative care nurses: A qualitative exploration, International Nursing Review (????). [11] H. S. Deore, M. Taranikanti, A. Gaur, S. Varatharajan, N. A. John, R. Katta, S. S. Taranikanti, M. Umesh, V. Ganji, K. Medala, Comprehensive, continuous, and compulsory monitoring of frailty in elderly, Journal of Family Medicine and Primary Care 12 (2023) 3194–3199. [12] D. J. McMaughan, O. Oloruntoba, M. L. Smith, Socioeconomic status and access to healthcare: interrelated drivers for healthy aging, Frontiers in public health 8 (2020) 231. [13] V. Haward, Impact of japan’s aging population on healthcare costs and the long-term care insurance system, Studies in Social Science & Humanities 3 (2024) 39–44. [14] L. Camaioni, P. Di Blasio, Psicologia dello sviluppo, Società editrice il Mulino, Spa, 2020. [15] E. Cambria, D. Das, S. Bandyopadhyay, A. Feraco, Afective computing and sentiment analysis, A practical guide to sentiment analysis (2017) 1–10. [16] J. Kim, J. Kim, D. Lee, K.-Y. Chung, Ontology driven interactive healthcare with wearable sensors,
Multimedia Tools and Applications 71 (2014) 827–841. [17] C. Villalonga, H. Pomares, I. Rojas, O. Banos, Mimu-wear: Ontology-based sensor selection for real-world wearable activity recognition, Neurocomputing 250 (2017) 76–100. [18] Blueprint personas, https://blueprint-personas.eu/, 2024. Accessed: 2024-11-14.