AI Strategy for Pharma
Clinical Trials & Commercial

Designed the study and protocol to benchmark three wearable sensors, including smartphone sensor data, against gold-standard clinical measures of disease progression, capturing step count and the intensity and frequency of ambulation in ALS patients across mobility levels, from independent walkers to cane users.

Outcome: Mapped which sensors and measures remain reliable as ambulation declines, giving the pharma client a defensible basis for device selection and a foundation for endpoint strategy in later trials.

Related Publications (Author & Advisor)

– Tracking ALS disease progression using passively collected smartphone sensor data (2024)

– Wearable device and smartphone data quantify ALS progression (2023)

Led the implementation and validation of wearable biosensor endpoints in a gene therapy trial for DMD, overseeing novel endpoint development, scientific validation against gold standards, protocol design, and data analytics methods.

Outcome: This advisory work gave the client a validated, defensible basis for having patients wear ankle sensors continuously for the full study, starting in Phase 2 and extended into Phase 3. The resulting data was processed and analyzed with advisory input, and is currently under review for publication.

Related Publications (Author & Advisor)

– One year activity monitoring outcomes in a phase 1b fordadistrogene movaparvovec study for DMD (under review, Npj Digital Medicine)

Designed a suite of novel digital assessments of fine motor function and manual dexterity (patent-pending) for patients with ALS. Led a patient advisory group that piloted the app, and authored the clinical study protocol for the resulting real-world evidence (RWE) study collecting multimodal data from ALS patients.

Outcome: A tablet-based, gamified assessment suite showing preliminary evidence of detecting fine-motor decline often missed in conventional clinical assessments. This advisory work gave the client a more sensitive, at-home measure of disease progression, with potential to adapt to other neurological diseases such as multiple sclerosis and Parkinson’s.

Led a patient advisory group to co-develop a companion app for patients with erythropoietic protoporphyria (EPP), a rare condition causing severe sensitivity to UV light. A wearable UV sensor syncs with the dedicated app to help patients monitor sun exposure, predict symptoms, and stay on track with their medication.

Outcome: This advisory work captured patient voice alongside SaMD regulatory requirements, building both into the product design from the start rather than retrofitting them after launch.

Contributed to a multi-site study validating a machine-learning toolkit that uses everyday smartphone typing patterns to identify early ALS symptoms and fine-motor decline.

Outcome: The tool separated slow from fast progressors and flagged subtle fine-motor decline earlier than the standard clinical assessment, suggesting the use of passive typing data as a complementary endpoint for the pharma client’s future trials.

Related Publications (Author & Advisor)

– A novel digital tool for detection and monitoring of ALS motor impairment and progression via keystroke dynamics (2024)

Validated objective speech biomarkers in ALS with bulbar onset, using voice samples patients record at home. Empirically compared two vendor platforms (a smartphone app and a web-based tool) to determine which produced acoustic data that better predicted ALS progression and correlated more strongly with gold-standard clinical assessments over time.

Outcome: Using machine learning models that predict slow vs. fast ALS progressors, the pharma client was able to strengthen trial stratification and refine inclusion/exclusion criteria.

Related Publications (Author & Advisor)

– Translating AI research into reality: summary of the 2025 voice AI Symposium and Hackathon (2026)