The accelerated aging of the global population has made Alzheimer’s disease one of the main public health concerns. According to projections, the number of people over the age of 60 will account for more than 40% of the total population by 2050, accompanied by a significant increase in cases of neurodegenerative diseases such as Alzheimer’s. Since current treatments are more effective when administered in the early stages of the disease, early diagnosis becomes a fundamental tool for slowing its progression and improving the quality of life of patients and their caregivers (Lopez‑de‑Ipina et al., 2021).
The Artificial Intelligence Platform for the Early Detection of Alzheimer’s Disease through Voice (IAEAV) has the primary objective of identifying patterns of cognitive decline based on the analysis of acoustic and linguistic features of human speech. This innovative tool builds upon previous studies showing that early neurological changes can manifest as alterations in language, such as reduced syntactic complexity, prolonged pauses, and grammatical errors (Fraser et al., 2016).
Using advanced natural language processing (NLP) techniques and deep learning, the platform aims not only to facilitate early diagnosis but also to provide a simple, low‑cost, easily accessible, and non‑invasive solution.
Voice data collection will be carried out through a simple and accessible mobile application that allows users to record their voices in various contexts, such as text reading, spontaneous narration, or responses to standardized questions. This application is designed for use both in clinical settings and at home, reducing access barriers and facilitating data acquisition in populations with limited resources.
The ease of use of the application will be key to ensuring the participation of a wide variety of users and to generating representative data for model training and validation (Lopez‑de‑Ipina et al., 2021).
The platform integrates a holistic approach that combines voice data collection with advanced analytical techniques. The recordings gathered through the mobile application are processed to extract acoustic features such as pitch, intensity, and pauses, as well as linguistic aspects such as semantic richness and verbal fluency errors (Toth et al., 2018). These features are then evaluated by deep learning models trained on representative datasets, enabling accurate and personalized detection.
One of the most relevant aspects of this approach is its accessibility. Unlike other diagnostic methods, such as neuroimaging or invasive biomarkers, voice analysis is non‑invasive and can be implemented in communities with limited resources. In addition, the platform includes a user‑friendly interface that allows healthcare professionals and caregivers to easily interpret the results, identify risk patterns, and monitor changes over time (Haider et al., 2022).
This technology not only seeks to improve clinical detection but also to contribute to scientific development through the generation of large volumes of voice data. These data can support deeper research into the relationship between linguistic alterations and neurodegenerative changes, fostering advances in the treatment and management of the disease (Fraser et al., 2016).
However, the project faces important challenges, such as ensuring data representativeness and complying with ethical and privacy regulations when handling sensitive information.
This platform offers an innovative and accessible approach to addressing one of the main public health challenges associated with population aging. The incorporation of a mobile application as the primary tool for data collection facilitates broader adoption and helps democratize access to early Alzheimer’s diagnosis, with the potential to significantly improve patient outcomes and reduce the economic and social impact of the disease in the coming decades.
In addition to the platform, a public dataset will be developed that includes voice recordings labeled and enriched with relevant metadata such as age, gender, educational level, and cognitive status of the participants. This dataset will be designed to ensure full anonymization and to comply with the highest ethical and regulatory standards for handling sensitive information.
The creation of this resource aims to foster collaboration and advance global research by enabling other artificial intelligence projects to use these samples to develop, validate, and compare models focused on Alzheimer’s diagnosis through voice.
The availability of this dataset will support greater methodological diversity and allow solutions to be adapted to different cultural and linguistic contexts, promoting open innovation that benefits the entire scientific and medical community.
