How a new AI system can help diagnose Alzheimer’s disease earlier and more accurately

Alzheimer’s disease is a progressive brain disorder that affects memory, thinking, and behavior. It is the most common cause of dementia, accounting for 60 to 80 percent of all cases1. According to the World Health Organization, around 50 million people worldwide have dementia, and this number is projected to reach 152 million by 20502. Alzheimer’s disease has no cure, and the current treatments can only slow down the symptoms, not stop the disease from worsening.

One of the biggest challenges in Alzheimer’s disease diagnosis is the lack of reliable and affordable biomarkers, which are measurable indicators of the presence or severity of a disease. Currently, the most accurate biomarkers for Alzheimer’s disease are obtained from cerebrospinal fluid (CSF) samples or brain scans, which are invasive, expensive, and not widely available3. Therefore, many people with Alzheimer’s disease are diagnosed at a late stage, when the brain damage is already irreversible.

However, a recent breakthrough in artificial intelligence (AI) could change this situation. A team of researchers from IBM and Pfizer have developed a new AI system that can predict the risk of developing Alzheimer’s disease by analysing a person’s speech. The system uses natural language processing (NLP) and machine learning (ML) techniques to extract features from the transcripts of a cognitive test called the Cookie Theft Task, which asks the participants to describe a picture of a boy, a girl, and a woman in a kitchen. The system then uses these features to classify the participants into two groups: those who are likely to develop Alzheimer’s disease within the next two years, and those who are not.

The researchers tested the system on a dataset of 703 participants from the Framingham Heart Study, a long-term study that tracks the health of thousands of people and their families since 1948. The system achieved an accuracy of 70 percent, which is comparable to the accuracy of CSF and brain scan biomarkers, but with much lower cost and invasiveness. The system also outperformed the accuracy of other speech-based biomarkers, such as word count, pauses, and repetitions.

The researchers believe that this AI system could be a game-changer for Alzheimer’s disease diagnosis, as it could enable early and accurate detection of the disease using a simple and accessible tool. The system could also help monitor the progression of the disease and the effectiveness of treatments over time. The researchers plan to further improve the system by incorporating more data and features, such as voice, tone, and emotion.

This AI system is an example of how AI can help solve some of the most pressing problems in health care and improve the quality of life for millions of people. AI is not only a powerful tool for creating and consuming content, but also a potential ally for fighting diseases and saving lives.