Natural language processing tool for cognitive scoring in Alzheimer's Disease patients

This technology is a natural language processing-based platform for estimating Mini-Mental State Examination (MMSE) scores from clinical notes, enabling cognitive assessment of Alzheimer’s Disease patients without additional testing.

Unmet Need: Reliable cognitive scoring from unstructured clinical data

Current methods for tracking cognitive decline in Alzheimer’s Disease rely on in-person administration of the Mini-Mental State Examination (MMSE) or extraction of structured electronic health record data, both of which are limited in availability and often incomplete. Clinical notes frequently contain relevant cognitive information, but this data is unstructured and difficult to analyze at a large scale. Although these notes are routinely collected in standard care, there are currently no tools that enable reliable, automated estimation of cognitive scores directly from unstructured clinical text.

The Technology: Automated cognitive scoring from routine clinical documentation

This technology is a natural language processing-based platform that estimates cognitive performance in Alzheimer’s Disease patients by analyzing unstructured clinical notes alongside basic patient demographic data. Natural language processing techniques are used to identify and quantify features relevant to cognitive status, which are then input into a machine learning model trained to predict Mini-Mental State Examination (MMSE) scores. This approach enables retrospective assessment of cognitive decline using routinely collected data without requiring additional clinical testing, allowing the model to generalize across varied clinical documentation styles and provider note formats.

This technology has been validated with >1,000 electronic health records from Alzheimer’s Disease patients.

Applications:

• Retrospective cognitive assessment
• Patient screening and stratification for clinical trials
• Clinical decision support for tracking cognitive decline
• Electronic health record (EHR)-based population health studies
• Evaluation of treatment effects on cognitive function
• Augmenting incomplete EHR datasets with estimated cognitive scores
• Longitudinal monitoring of cognitive changes in routine care
• Identifying early cognitive decline
• Integrating cognitive metrics into predictive models for disease progression

Advantages:

• Enables automated estimation of cognitive scores from existing clinical notes
• Reduces reliance on structured EHR data or in-person Mini-Mental State Examination (MMSE) administration
• Improves cognitive monitoring workflows
• Supports consistent cognitive assessment
• Cost-effective

Lead Inventor:

Chunhua Weng, Ph.D.

Related Publications:

• Idnay B, Zhang G, Chen F, Ta CN, Schelke MW, Marder K, Weng C. “Mini-mental status examination phenotyping for Alzheimer's disease patients using both structured and narrative electronic health record features.” J Am Med Inform Assoc. 2025 Jan 1; 32(1): 119-128.

Tech Ventures Reference:

• IR CU25391

• Licensing Contact: Joan Martinez