In our previous posts, we talked about how we’re looking to use language as a method for treating neurodegenerative diseases. Most recently, we discussed using technology to create a tool that provides language-based exercises to both assess risk and provide early treatment for neurodegenerative diseases. We discussed our ideas with Dr. Murray Grossman, an expert in language and neurodegenerative diseases at Penn Medicine. He provided us with some very useful feedback that has helped us refine our search for a solution.

Dr. Grossman informed us that Primary Progressive Aphasia (PPA)—a form of aphasia that often accompanies Alzheimer’s Disease or Frontotemporal Dementia—actually produces very early linguistic indicators. These indicators are very subtle (cannot be heard with the human ear) and take the form of changes in the word choice and acoustic profiles of a patient’s speech. However, they are detectable enough that they can be picked up in speech over the phone, which suggests that an app that a patient runs on a mobile device could potentially pick up these markers. Based on this information, we’ve decided to make our focus neurodegenerative diseases associated with Primary Progressive Aphasia, and we plan to find a method for automatically detecting early linguistic indicators of these diseases from within a mobile app.

From our discussion with Dr. Grossman, we also learned that the primary cause of PPA is the presence of defective Tau proteins within a patient’s cerebrospinal fluid. He spoke about treatments that are being evaluated through experimental trials, which involve biologically-based removal of these proteins. Results from this testing should become available within the next few years. Dr. Grossman also suggested that linguistic exercises are likely to be insufficient for fully treating these diseases. Based on this information, we’ve decided to pivot away from building a treatment tool for neurodegenerative disorders, as it seems that most prospects for treatment require medical intervention to be effective. Instead, we plan to focus on early diagnosis, which may enable patients to make use of these new treatment methods before their neurodegeneration becomes significant.

We also learned from Dr. Grossman much about the different variants of PPA. While much of how speech is impacted in the different forms of PPA is significantly different, in particular semantic variation PPA (svPPA) compared to nonfluent/agrammatic PPA (naPPA), much of the same early detection can be done for all forms of the disease. Based on this information, we have decided to make our automated detection app for all variants of PPA. While we had known that there were several variants of PPA prior to our meeting with Dr. Grossman, his insight helped us to decide to create our automated detection for all variants instead of either making it for one variant or making different detection tests for each variant, both of which were methods we had been previously considering.