The design of our platform ecosystem includes a neuromodulation system and data collection devices to be installed in a person’s home. It also incorporates cloud storage for long-term data storage. A VPN network enables remote updates to aDBS algorithms, and it allows data to be securely transmitted to researchers and clinicians.

Where is the Person in Personalized Medicine? The Missing Expert in Adaptive Neurotechnology.

My Ph.D. dissertation presents a case study on delivering adaptive Deep Brain Stimulation (aDBS) therapy for Parkinson’s disease (PD) in the comfort of peoples’ homes. Delivering this promising therapy outside clinical settings requires remotely maintainable technical solutions and expertise from diverse fields, including machine learning, clinical neurology, and data science.

I designed and deployed a platform ecosystem to remotely collect multi-modal data from the home of a person with PD for 2 years, and I contributed novel methods for remotely evaluating aDBS outcomes in the first study to remotely evaluate therapeutic outcomes using multi-view video cameras, wearable sensors, neural signals, and participant-submitted feedback.

Importantly, I analyzed the experience of the research participant and demonstrated that they are an invaluable domain Expert whose input will lead to more viable and human-centered neurotechnologies. My analysis further showed that the expertise of research participants is frequently missing from neurotechnology research spaces, and I call on all neurotechnology researchers to include the people that their technologies are designed for into the research process. Drawing on practices from Participatory Action Research, I proposed a flexible framework for how neurotechnologists can co-research with participants.

Listen to my Ph.D. defense talk here, get a copy of my slides here, and read my dissertation here.

I measured various aspects of movement quality while a research participant performed finger tapping, opening and closing a fist, and wrist rotation tasks in their home. By comparing measurements when the participant received different levels of stimulation amplitude, we can continuously assess how well symptoms are treated.

In-Home Video and IMU Kinematics of Self Guided Tasks Correlate with Clinical Bradykinesia Scores

I developed metrics to remotely assess the symptom severity for people with Parkinson’s disease using 2D pose estimates from videos and acceleration and velocity data from smartwatches. By correlating metrics like speed and range of motion with clinical UPDRS scores, I demonstrated that remote assessments could match gold-standard clinical evaluations.

This was the first study to remotely evaluate deep brain stimulation (DBS) therapy outcomes using such diverse data sources including multi-view video cameras. This approach enabled us to identify clear effects of stimulation amplitude on movement quality, providing insights into optimizing adaptive DBS algorithms. I applied these same methods in a later study evaluating a novel movement-triggered aDBS algorithm.

A Participant’s Experience Participating in an At-Home Pilot Study: A Reflexive Thematic Analysis

Research on the experiences of research participants exists, however it is frequently published in separate places from the research that impacts their lives, such as clinical neurology. This is notable because neurological diseases impact each person uniquely and subjectively, often in ways we cannot ascertain without directly asking the person.

I designed a reflexive thematic analysis of the experience of a research participant in my prior work investigating adaptive Deep Brain Stimulation at home for people with Parkinson’s disease. I developed several themes demonstrating that the fields of clinical neurology and neural engineering stand to benefit immensely from co-researching with participants in a collaborative approach. For instance, the participant was uncertain about navigating trade-offs in their movement abilities while performing some self-guided experimental tasks at home, which introduced noise into the collected data. These kinds of issues can be mitigated by co-designing experiments with the participant.

My analysis demonstrated the value of integrating the participant’s domain expertise into neurotechnology research and design. The full analysis, as well as actionable insights for improving our neurotechnologies via co-design, can be found in Chapter 4 of my dissertation.

Clinical neurology citation counts from the biomedical database PubMed that include the term “patient experience” compared to those that do not.
Created via Ed Sperr’s excellent visualization tool here.