Data Driven Diagnosis of Alzheimer's Disease
summary
Scraped Alzheimer Society UK forums yielding 3600 threads and 78,000 posts, cleaned with BeautifulSoup and aggregated by user; features include readability-lxml scores (Dale‑Chall), TextBlob sentiment, POS ratios, and cosine/edit distances between consecutive posts to catch repetition. Models use scikit‑learn random forests and SVM (RBF); user‑level median aggregation gives ~63% accuracy while balanced post‑level sampling reaches 68% per-post accuracy. Memory/change features and post similarity drive most signal; dataset bias, coarse labeling and lack of tuning mean this is exploratory not clinically deployable.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:09Alzheimer's impact
- 3:16Team and scope
- 5:14Symptoms and progression
- 9:44Data from forums
- 13:21User labeling strategy
- 17:00Feature engineering
- 22:44Initial modeling
- 26:04Classification results
- 27:30Future directions
- 28:13Summary and collaboration
- 28:58Q&A
related talks
27mMachine Learning Methods in Detecting Alzheimer’s Disease from Speech and Language
32mAnalyzing social media for suicide risk using natural language processing
32mTriaging Feedback Form Data
37mIs having dementia linked to where you live?
01h 24mIntroductory tutorial on data exploration and statistical models
18mMultimodal Classification of Moderated Online Pro-Eating Disorder Content
38mEthical Artificial Intelligence to Advance Welcoming Communities and Identify Potential Communica...
06mLeveraging Collaborative-Filtering for Personalized Behavior Modeling: A Case Study of ...