The Utilization of AI-Based Chatbot for Early Detection of Depression in the Elderly

Background: Late-life depression is often misdiagnosed due to multiple factors such as social isolation, cognitive decline, and the onset of aging symptoms. Therefore, early identification of depression plays a significant role in improving outcomes and preventing deterioration in the mental health of older adults. The aim of this study was to examine the potential of AI-based Chatbot, utilizing Natural Language Processing (NLP), for the early detection of depression symptoms among the elderly. Materials and Methods: In an experimental study, elderly individuals were introduced to an AI Chatbot for depression screening. This Chatbot was made available through mobile phones and tablets during routine health assessments. The Chatbot used the PHQ-۹ questionnaire (Patient Health Questionnaire-۹) to evaluate depression symptoms based on responses to a series of structured and semi-structured questions about mood, energy level, and interest in daily activities. Natural language processing algorithms analyzed the responses for indicators of sadness, hopelessness, irritability, and withdrawal. Additionally, in the sentiment analysis model, the tone and word choices in the responses of the elderly were evaluated to identify potential changes in mood. Follow-up questions were activated based on initial responses, and changes in behavior among the elderly were monitored over time. If the Chatbot predicted a higher likelihood of depression, it would notify primary care specialists for further evaluation. Results: The AI Chatbot demonstrated high sensitivity in detecting early signs of depression, particularly among elderly individuals who were reluctant to report their symptoms during in-person consultations. The Chatbot was able to provide continuous and real-time evaluations, delivering results more quickly than traditional screening methods. Depression symptoms were detected in ۳۰% of the elderly users of the Chatbot, which had gone unrecognized through conventional methods. These findings confirm the potential of AI-based Chatbot for early detection of depression, allowing for timely interventions and improved care. Conclusion: AI Chatbot that utilize Natural Language Processing and sentiment analysis provide valuable and accessible tools for the early identification of depression in older adults. Although challenges such as ensuring diagnostic accuracy and maintaining data privacy remain, the potential of AI-based tools to enhance the screening and management process of depression among the elderly is critically important.

کلیدواژه ها:

Depression ، Elderly ، Chatbot ، Artificial Intelligence ، Natural Language Processing ، Sentiment Analysis