About

Alzheimer’s disease is a progressive condition that affects millions globally, leading to memory loss, cognitive decline, and emotional isolation. In Kuwait, the growing prevalence of Alzheimer’s highlights the urgent need for innovative and culturally sensitive care solutions. Most assistive technologies today lack adaptation to local dialects and traditions, making them less effective in engaging Arab patients. To address this, we introduced Robot-based Framework for Helping Alzheimer's Patients and developed the Alzheimer Care Companion Robot, an AI-powered social robot designed to interact with patients using the Arab-Kuwaiti dialect. It includes 271 structured conversations across active, proactive, and Graphical User Interface (GUI)-based formats, focusing on familiar topics such as family, memories, and daily life. It uses personalized images and videos to support memory recall and emotional engagement. This initiative enhances patient well-being, reduces caregiver burden, and promotes culturally grounded care. It also opens pathways for inclusive AI development in the Arab world, offering scalable, empathetic support to improve quality of life.

Description

As memory fades, the ability to connect with personal identity, relationships, and surroundings declines, which makes Alzheimer’s one of the most devastating global public health concerns. In 2020, the World Health Organization (WHO) reported that Alzheimer's and dementia-related deaths in Kuwait accounted for 139 cases, representing approximately 1.82% of the total deaths recorded. Furthermore, the age-adjusted death rate for Alzheimer's and dementia in Kuwait was 17.42 per 100,000 population, positioning Kuwait at the 96th rank worldwide in terms of prevalence. In recent years, developers have been exploring innovative technological solutions to aid Alzheimer's patients. These solutions include tools for Alzheimer's detection, live conversation assistance, and memory support. One particularly creative tool that has emerged is the use of talking robots. These robots can be programmed to engage in conversations with Alzheimer's patients, assisting them in recalling memories and refreshing their cognitive abilities.

This project initiative introduced the Alzheimer Care Companion Robot, a culturally adaptive, AI-powered robot designed specifically to engage Alzheimer’s patients through natural language interactions in the Arab-Kuwaiti dialect. The solution aims to support memory recall and emotional connection by initiating meaningful conversations and displaying personalized visual content such as images and videos related to the patient’s life experiences. A structured framework was developed for creating conversational agents tailored to the cognitive and emotional needs of Alzheimer’s patients. It provides a categorization of 271 interactive dialogues. Samples are illustrated in TABLE 1. Three main categories of conversations are classified: 1) active conversations which consist of 112 different dialogs; 2) proactive conversations which consist of 109 different conversations; and 3) Graphical User Interface (GUI) interaction which consists of 50 different conversations.

TABLE 1: Samples of conversations.

By combining localized language use, culturally familiar content, and assistive robotics, this project contributes to the growing body of knowledge on human-centred AI in healthcare and offers a novel approach to personalized, compassionate Alzheimer’s care in the Arab world.

Outcomes

Robot-based Framework for Helping Alzheimer's Patients

The framework consists of four main phases, as depicted in Figure 1: studying and data collection phase (SDCP), system requirement phase (SRP), conversation preparation phase (CPP), and implementation phase (IP). Each phase consists of several steps and performs specific tasks.

• • Studying and Data Collection Phase (SDCP): the phase of understanding Alzheimer's disease and the Alzheimer's patient's needs. SDCP consists three stages, expert interviews, Alzheimer’s patient interview, and data collection stage.
• • System Requirement Phase (SRP): the phase to define functional and non-functional requirements, the hardware, and the software of the system.
• • Conversation Preparation Phase (CPP):the phase of collecting and preparing a dataset and conversations of Alzheimer’s patients such as family, relationships, hobbies, etc.
• • Implementation Phase (IP): the phase of specifying and defining the system’s components, the inputs, outputs, and system interaction units.

Figure 1: Framework architecture.

The Proposed Alzheimer Care Companion Robot

The framework incorporates three different components: inputs, outputs, and system interaction unit to engage with patients and aid memory recall and is illustrated in Figure 2.

Figure 2: The proposed solution (ACCR).

Input Components

• • Camera: The camera plays a crucial role in detecting Alzheimer's patients when they approach the robot. It captures visual data that is used by the Face Recognition Model (FRM) to identify and differentiate patients from other users.
• • Microphone: The microphone records speech from the patients as they engage in conversations with the robot. The Speech Recognition Model (SRM) processes the audio input and converts it into text for understanding patient's spoken commands and queries.
• • Touch Screen: The touch screen allows patients to interact directly with the robot's graphical user interface. It enables a user-friendly and intuitive way for patients to communicate and engage with the robot.

Output Components

• • Speaker: The speaker is responsible for delivering audio output from the robot.
• • Behaviors: These behaviors are designed to create a comforting and non-intrusive environment.
• • Screen: The built-in screen on the robot displays relevant images and videos to aid memory recall in Alzheimer's patients during conversations. This feature enhances the patient's engagement and stimulates reminiscence.

System Interaction Unit

• • Proactive Conversations Model (PCM): The PCM enables the robot to initiate conversations with Alzheimer's patients proactively.
• • Active Conversations Model (ACM): In the ACM, the patient asks questions and the robot answers them. The robot generates appropriate responses to the patient's queries and maintains the flow of conversation throughout the interaction
• • Speech Recognition Model (SRM): The SRM is responsible for converting the patient's spoken language (Arab-Kuwaiti dialect) into text data.
• • Face Recognition Model (FRM): The Face Recognition Model (FRM) analyzes visual data from the camera input to identify and distinguish Alzheimer's patients from other users. This enables personalized interactions with the patients and enhances their experience.
• • Graphical User Interface (GUI): The GUI is the interface displayed on the robot's touch screen. It enables patients to interact with the robot, view relevant images, and trigger conversations on various topics like family, relationships, and daily activities.

Experimental Results

The project utilized the Pepper robot developed by SoftBank Robotics Corp to implement the solution. The team conducted tens of scenarios to test the proposed solution using the proactive conversation, the active conversation, and GUI. Figure 3 illustrates an example of the interaction with the robot. It was very interesting to the individual, that there is a robot speaking and interacting with them. The person was very happy and excited about this experience.

Figure 3: A real experimental scenario of the proposed solution.

Scenario 1: Proactive Conversation

In this scenario, the robot initiates the conversations with the patient on a range of topics, such as family, relationships, hobbies, and daily activities. This can help to trigger memories and provide a sense of enjoyment for the patient. The robot also displayed pictures of the patient's favorite places, such as their hometown (Figure 4-a), their close friends (Figure 4-b), or ask the patient to describe their experiences.

a) The Alzheimer's preferred place.

b) Friends for one Alzheimer's patient case.

Scenario 2: Active Conversation

In these scenarios, the patient asks questions, and the robot replays them. For example, the patient can ask the robot about his family, friends, habits, prayer time, or medical time. The patient may ask questions such as "Who am I?" and "What is my wife's name?" and the robot provides answers such as "You are Dhari" and "Your wife is Fatima".

Scenario 3: Graphical User Interface

The proposed GUI of the robot has four main categories or buttons displayed on the screen: 1) Family and Relationships (أهلك و ربعك); 2) Memories (اللي ماله اول ماله تالي); 3) Your Health (صحتك); and 4) Prayer Times (اوقات الصلاة). When the user selects Family and Relationships (أهلك و ربعك), the robot will display on its screen another window (Figure 5-a), that allows the user to select either family or relationships button. If the relationships button is selected, the robot will display on its screen the slideshow displaying a picture of the user’s friends and will talk about those friends (Figure 5-b).

a) Graphical User Interface

b) Photo of user’s friends

Figure 5: Examples of a real GUI Scenarios.