The connection between neurodegenerative brain diseases, abnormalities in the brain tissue and the associated symptoms is often unclear; for example, some symptoms are common to multiple conditions, and the clinical picture can vary widely from patient to patient, even resulting in misdiagnoses. Researchers and doctors would like more clarity about this.
Could AI provide good services in that regard? Dutch researchers thought so, and they thought well: Nienke Mekkes et al. present their findings in Nature Medicine. They took advantage of the fact that all post-mortem brain tissue stored in the Dutch Brain Bank is accompanied by an accurate description of the neuropathological diagnosis and the course of the disease, including the symptoms.
Incorrectly diagnosed
Mekkes et al. ‘feed’ an AI language model with that combination of texts and tissue characteristics, belonging to 3042 donors. This resulted in ninety different symptoms spread over five different domains: psychiatric symptoms (such as depression, psychoses), cognitive symptoms (such as dementia, memory problems), motor problems (such as tremors) and sensory symptoms (including feeling things that are not there).
A second AI prediction model was then developed, which should make diagnoses based solely on the clinical picture. This turned out to work quite accurately, except for rare disorders. And thanks to this model, the researchers found something else: a group of donors who had apparently been incorrectly diagnosed while alive. It appears, according to one of the researchers in an explanation to the press, that a group of people with Alzheimer’s disease exhibit symptoms that are more reminiscent of Parkinson’s disease. And that in another group of people frontotemporal dementia manifests itself as Alzheimer’s disease.
Unique resource
Mekkes et al. believe that they have created ‘a very unique resource that could benefit a wide range of researchers’. They mention: epidemiologists who study the (temporal) symptomatology of various brain disorders, molecular biologists who aim to gain a deeper understanding of the cellular and molecular features that give rise to neurodegenerative diseases and computational researchers who want to build predictive models for diagnosis and prognosis of patients with dementia.
Also read
2024-03-14 15:45:00
#connection #symptoms #brain #tissue #abnormalities
