Voices
Imagine, you come to the oncologist with a stomach ache. After asking a few questions, he concludes that it is most likely cancer and recommends chemotherapy. A radical diagnosis and treatment without a puncture or MRI scan. Unbelievable? Unfortunately, this is the reality in psychiatry. As a psychiatrist I have no objective diagnostic test at hand, I have to decide for myself whether someone has a disorder or not. That decision is subjective and unreliable, but it has major consequences for many people: because 4 out of 10 Dutch people will develop a psychiatric disorder at some point.
Fortunately, there is a solution. As a linguist, I know that speech carries an enormous amount of information about how someone is doing. Not only what is said, but also how it is said is important. We have developed a technical application with which we can reliably analyze whether someone has a psychosis or not based on their speech. We use the latest developments in machine learning and natural language processing (NLP), such as predecessors of ChatGPT. In addition, we have shown that these techniques provide clinically relevant insights in the study of auditory verbal hallucinations (hearing voices). In the future, these analyzes may be used to detect a relapse in psychosis earlier, so that you can see a psychosis coming before people become seriously ill.