Work

AI-powered speech assessment for early detection of neurodegenerative diseases.

Business Challenge

Boehringer wanted to digitize an 8-step speech assessment protocol (originally from Boston University) used to pre-diagnose neurodegenerative diseases such as Alzheimer’s.

The manual process was slow, inconsistent, and required trained staff — which limited scale. Elderly patients often struggled with complex instructions, and researchers needed a secure way to manage and share sensitive data.

Approach

I designed a highly simplified interface that allowed elderly patients to complete assessments independently, following clear on-screen instructions.

We integrated an AI speech recognition engine into Boehringer’s infrastructure, enabling automated analysis and consistent data capture.

On the research side, I designed a secure backend for clinicians to monitor assessments, compare results, and manage high-confidentiality datasets in a compliant environment.

I also coordinated with multiple clinical centers in Germany, training staff and ensuring the system worked in real-world conditions — not just in theory.

Outcome

• Turned a multi-day manual process into a scalable digital assessment
• Improved accessibility for elderly patients
• Enabled secure, compliant sharing of speech-based biomarkers
• Created the foundation for future clinical trials using speech as a diagnostic indicator

Assessment screens remain confidential due to NDA — which is fine, because pharma secrets are serious business.