AI in healthcare - Deep learning for precision psychiatry
summary
Company develops a deep learning pipeline called Vulkan framework that ingests pooled clinical trial data and compresses 100,221 candidate predictors into actionable features, uses bootstrap k‑fold validation and a naive per‑patient counterfactual loop to rank four medications, and deploys results in a physician‑patient app combining coded best‑practice decision trees with AI treatment probabilities. Models report AUC 0.69, a naive 32% relative remission boost (34% → 46%) and a conservative 7.2% relative improvement with p = 0.01.
AI-written summary · about
chapters (26)
- 0:00Introduction
- 2:33Trials and Error
- 4:51What is Deep Learning
- 5:41Image Recognition
- 5:53Deep Learning for Psychiatry
- 7:10Traditional Machine Learning
- 9:19Vulkan
- 9:53Pipeline
- 10:59Naive differential analysis
- 12:04Bootstrap Kfull validation
- 13:04Results
- 15:49Clinical algorithm
- 17:10Treatment selection page
- 17:59Thank you
- 18:33Question from Jasper
- 18:54Questionnaires
- 19:39Drugfree treatments
- 20:04Data integration
- 21:11The use of AI
- 21:31Impact
- 22:39Current models
- 23:27Recruitment
- 24:17Missing data
- 24:34Data amputation
- 25:21Brain scans
- 27:29Outro
related talks
25mAI for healthcare: Scaling Access and Quality of Care for Everyone
37mAI in medicine: where do we stand?
44mAI in the Clinic: Prediction is Better Than Cure
23mAI in the Life Science Sector
36mAI in Healthcare
47mAI for making decisions in clinical medicine
47mAI for the Proper Use of Medications
17mAI Innovation in Health