The applications of Artificial Intelligence in the healthcare industry are indeed laudable. Somewhere or the other it keeps on fascinating all with its mind-blowing potential. Recently a collaborative team of IBM and Pfizer has accomplished an AI system to predict the risks of Alzheimer’s disease at the early stage itself.
The research team has conducted a study on how Artificial Intelligence will diagnose Alzheimer’s disease by analyzing linguistic patterns and writing. Generally, the prediction will be done based on the data of Brain scans and other clinical test results.
But this new AI model is trained to analyze data from the past three generations. It uses historical data and information from the “Framingham Heart Study” which helps to predict among the diverse environments. AI will be trained in all aspects to diagnose the signs several years before, in which the existing system lags.
The AI model will help practitioners in detecting meticulous changes at the earlier stages. This will further help doctors to decide whether further examinations are required to be carried out. The researchers have trained the AI model with a digital version of the handwritten responses of the participants.
The study was carried out with the participants from the Framingham Heart Study. They were given a scenario and asked to reproduce it in their own words. The participants were asked to describe a woman who is doing dishes lost in some thoughts and her two kids are trying to raid a cookie jar without her knowledge.
The responses are analyzed and the detection process is carried out with the help of spelling errors, frequent usage of the same words, lack of grammatical sentences, use of simple sentences. The AI model considers these parameters while detecting the early signs of Alzheimer’s.
In general, the AI model is designed in such a way to detect cognitive impairment by linguistic features associated with them.
The study identifies that the new AI model is capable of detecting early symptoms of Alzheimer which accounts for a 70% accuracy rate. The accuracy can be further be enhanced by integrating the handwriting as well. Currently, the AI model reads the digital transcripts of the handwritten responses for analysis.
