Using artificial intelligence to process natural language, a research group assessed speech among individuals with Parkinson’s disease. AI analysis determined that these patients used a higher number of verbs and fewer nouns and fillers.
The study was led by Professor Masahisa Katsuno and Dr. Katsunori Yokoi from Nagoya University Graduate School of Medicine, in collaboration with Aichi Prefectural University and Toyohashi University of Technology. They published their findings in the journal Parkinsonism & Related Disorders.
Natural language processing (NLP) technology is a branch of AI that focuses on helping computers understand and interpret substantial quantities of human language data by using statistical models to find patterns. Given that patients with PD encounter a range of speech-related difficulties, including impaired speech production and language usage, the group used NLP to analyse differences in the speech patterns of patients based on 37 characteristics using transcripts derived from spontaneous conversations.
The analysis unveiled that patients with PD employed fewer common nouns, proper nouns, and fillers per sentence. Conversely, they exhibited a higher percentage of verbs and variance in case particles — an important feature of the Japanese language — per sentence.
According to Yokoi, “When I asked them to talk about their day in the morning, a PD patient might say something like the following, for example: ‘I woke up at 4:50 am. I thought it was a bit early, but I got up. It took me about half an hour to go to the toilet, so I washed up and got dressed around 5.30 am. My husband cooked breakfast. I had breakfast after 6 am. Then I brushed my teeth and got ready to go out.'”
Yokoi continued, “In contrast, someone from the healthy control group might say something like this: ‘Well, in the morning, I woke up at six o’clock, and got dressed, and, yeah, washed my face. Then, I fed my cat and dog. My daughter prepared a meal, but I told her I couldn’t eat, and I, umm, drank some water.'”
“While these examples were created to exemplify the conversational characteristics of individuals with PD and healthy individuals, it is worth noting that the overall length is comparable,” explained Yokoi. “However, PD patients employ shorter sentences compared to those in the control group, resulting in a greater number of verbs in the machine learning analysis. The healthy control group also uses more fillers, such as ‘well’ or ‘umm’, to link sentences.”
The most promising aspect of this research is that the team conducted the experiment on patients who had not yet displayed the characteristic cognitive decline associated with PD. Consequently, their findings offer a potential means of early detection to distinguish PD patients.
“Our results suggest that even in the absence of cognitive decline, the conversations of patients with PD differed from those of healthy subjects,” concludes Professor Katsuno, the lead researcher. “When we attempted to identify PD patients or healthy controls based on these conversational changes, we achieved over 80% precision in identifying PD patients. This outcome suggests the possibility of diagnosing PD through language analysis using natural language processing.”
