On the Potential of Using Artificial Intelligence to Handle Healthcare Instructions for the Elderly

Abstract

Telehealth services have been playing an increasing role in the communication and management of healthcare. Cognitive decline in the elderly presents challenges for the effectiveness of these services. This research shares findings from the development of a smartphone app prototype that can reduce select burdens on the growing population of the aging-in-place elderly, with normal-aging cognition or with mild cognitive impairment (MCI), when receiving healthcare instructions. These burdens include comprehending and summarizing instructions, extracting actions, and creating planning reminders. The app accepts audio input representing instructions spoken by healthcare professionals. After performing speech-to-text transcription, the app applies artificial intelligence (AI) classification, natural language processing and summarization capabilities to automatically produce action reminders. Audio samples varying in complexity (number and order of actions) and speech speed are used to test the app. This categorization is inspired by the established distinct deterioration of executive functions (e.g. planning, working memory, and sustained attention) and temporal information processing that contribute to cognitive aging. The app’s processing speed as well as its accuracy and completeness of automatic creation of action reminders are measured. When compared to several widely available classic and AI-powered assistants of similar or partial functionality in smartphones, the results demonstrated notable deficiencies in existing assistants. This includes a decreasing probability of capturing three or more actions and an increased probability of replays for larger audio samples. Conclusions from this research call for an increased attention in the commercial and scientific domains to combat cognitive challenges affecting the elderly’s handling of healthcare instructions.